fredmiranda.com Login
This topic is in the 2021 archive. You can read it and like its posts here. It is read only.

Archive 2021Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount

  
 
bjornthun
Offline
Details
p.8 #1Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
philip_pj wrote:
Thanks for the info, Nikon is incomprehensible to any outsider. Being at the other end of the spectrum from Sigma's concept of modern photography, I'll check out the Novoflex for a 'dumb' adapter. Don't care about exif after all these happy years of CY/R/M lenses on Sony. Suspiciously, no one lists their adapter weights. Just need to be able to set the aperture by hand and I'm fine.


I’d choose an adapter of high mechanical quality with accurate length to maintain accurate film to flange distance of the lens, to ensure that Nikon wide angles with CRC perform optimally. CRC = close range correction, which is Nikon’s term for FLE or floating elements. This applies also F-mount Zeiss and Voigtländer lenses with floating elements.
Oct 16, 2021 at 07:31 PM
DavidBM
Offline
Details
p.8 #2Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
Fred Miranda wrote:
DavidBM wrote
Fred Miranda wrote
DavidBM wrote
Fred Miranda wrote
DavidBM wroteHow’s the sigma shaping up?
I did a quick test today right after it arrived. It's flat field and sharp from wide open to the very corners. It's not capable of resolving high frequency detail like the Leica 90mm f/2 APO but it's not that far behind.

This was surprising to me as I saw zero distortion:
https://www.fredmiranda.com/forum/topic/1715801/25#15736675
So it must have the non discretionary manufacturer profile turned on for distortion. More usual for just LaCA. In a way that’s a good sign….that sharpness across the field is after correction! Promising!

Some Sony lenses for APS have had non-discretionary manufacturer profiles in LR for distortion if I recall….maybe also the 2/28? But I’m not sure…
This is the first time I see distortion included in a LR built-in profile. With Sony full frame lenses, there is only automated correction for "lateral CA" and that includes the Sony 28mm f/2.

Yes, both 24mm and 90mm DG DN are sharp to the edges and improve slightly stopping down. They are not the sharpest I've seen but perform at very high level. The 23-year-old Leica APO-Summicron-M 90 f/2 ASPH @f/2.8 beats it with only 5 elements. However, when comparing both lenses wide open (f/2 vs f/2.8), they are very similar at infinity.

Also similar stopped down to At f/4-5.6 across the field. The Leica is still a tad better even at f/8 at pixel level but it would be hard to differentiate them in real world images. Great value for a well-built $650 compact lens with AF!

All this with distortion correction already applied for the Sigma.
Ah, sorry, must have misremembered about compulsory distortion correction on the 2/28! It needed it!
This is OT for this thread but I'm shocked by the level of distortion from the new 24/2 DG DN and 90/2.8 lenses.
However, this could be a blessing in disguise. Read on...

Whenever the correction is "barrel", it requires image cropping and in the case of the Sigma 24/2 and Sony 28/2 lenses, it requires a lot of cropping. The difference between the two is that the Sigma is uber sharp before "distortion correction + crop" and the Sony is not.

So, what Sigma did was to basically design a 22mm FL lens and let distortion go crazy relying on auto-correction
...Show more →

Exactly. Allowing distortion allows the lens to be incredibly sharp uncorrected, and possibly therefore sharper even resized and *corrected* that you could have done for the same price and size if designed without distortion. As you say, the 2/28 was not incredibly sharp even before correction, and so after correction…

PS is it a crop on the Leica DNG or a crop and resize? Ie are the pixel dimensions of the corrected file the same as usual for the Leica files?
Oct 16, 2021 at 07:38 PM
Fred Miranda
Offline
Admin
Upload & Sell: On
Details
p.8 #3Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
DavidBM wrote:
Fred Miranda wrote
DavidBM wrote
Fred Miranda wrote
DavidBM wrote
Fred Miranda wroteI did a quick test today right after it arrived. It's flat field and sharp from wide open to the very corners. It's not capable of resolving high frequency detail like the Leica 90mm f/2 APO but it's not that far behind.

This was surprising to me as I saw zero distortion:
https://www.fredmiranda.com/forum/topic/1715801/25#15736675
So it must have the non discretionary manufacturer profile turned on for distortion. More usual for just LaCA. In a way that’s a good sign….that sharpness across the field is after correction! Promising!

Some Sony lenses for APS have had non-discretionary manufacturer profiles in LR for distortion if I recall….maybe also the 2/28? But I’m not sure…
This is the first time I see distortion included in a LR built-in profile. With Sony full frame lenses, there is only automated correction for "lateral CA" and that includes the Sony 28mm f/2.

Yes, both 24mm and 90mm DG DN are sharp to the edges and improve slightly stopping down. They are not the sharpest I've seen but perform at very high level. The 23-year-old Leica APO-Summicron-M 90 f/2 ASPH @f/2.8 beats it with only 5 elements. However, when comparing both lenses wide open (f/2 vs f/2.8), they are very similar at infinity.

Also similar stopped down to At f/4-5.6 across the field. The Leica is still a tad better even at f/8 at pixel level but it would be hard to differentiate them in real world images. Great value for a well-built $650 compact lens with AF!

All this with distortion correction already applied for the Sigma.
Ah, sorry, must have misremembered about compulsory distortion correction on the 2/28! It needed it!
This is OT for this thread but I'm shocked by the level of distortion from the new 24/2 DG DN and 90/2.8 lenses.
However, this could be a blessing in disguise. Read on...

Whenever the correction is "barrel", it requires image cropping and in the case of the Sigma 24/2 and Sony 28/2 lenses, it requires a lot of cropping. The difference between the two is that the Sigma is uber sharp before "distortion correction + crop" and the Sony is not.

So, what Sigma did was to basically design a 22mm FL lens and let distortion go crazy relying on auto-correction in-camera to remove the barrel distortion and match the 24mm FL. However, in the process, a sharp and usable section of the image is set to be cropped to allow for the big barrel correction but some images do not require distortion correction and therefore could benefit from that extra range. It's almost like having a 21mm lens with high distortion and a 24mm lens with no distortion in one. The former is actually sharper at the very corners!

This is all hidden and done behind the scenes for the Leica SL2 and I have not checked the Sony yet. On the Leica, there is no way to turn distortion correction off in-camera or in-post (with LR or CaptureOne). So, to gather this information, I used the "DNG Cleaner" app which strips "cooked" lens correction + crop from DNG files, which in turn, allows me to see the real uncorrected RAW image.
Exactly. Allowing distortion allows the lens to be incredibly sharp uncorrected, and possibly therefore sharper even resized and *corrected* that you could have done for the same price and size if designed without distortion. As you say, the 2/28 was not incredibly sharp even before correction, and so after correction…

PS is it a crop on the Leica DNG or a crop and resize? Ie are the pixel dimensions of the corrected file the same as usual for the Leica files?


Here is an image showing why there is an actual crop with resize after the massive "barrel" distortion correction for the Sigma 24/2 DG DN. Look at the "white area" of the image (second set below). That area must be cropped out of the image after correction which in turn means less pixels afterwards.

So the image goes from 47.3 MP to 46.7 MP after correction.

BTW: This correction is at +29 (!!) in Lightroom. For the Sigma 90/2.8, its pincushion is not as severe at around: -13 which still way higher than average. The good news for the 90/2.8 is that since there is minimal crop and the correction is not as massive, loss of high frequency detail is minimal at the very corners. (I checked)

What I find interesting is that Sony users don't get the baked "distortion" correction in LR but Leica users do.
So L-mount shooters never get to see the uncorrected version unless they use different post-processing software apps like DxO PureRAW, RAWDigger, Iridient, DNG Cleaner, etc..





Corrected (TOP) | Uncorrected (BOTTOM) - fisheye territory? :)






Corrected (TOP) | Manually corrected before crop (BOTTOM) - Now you can see why I said it could also be used as a high distorted ~22mm lens!

Oct 16, 2021 at 08:04 PM
freaklikeme
Online
Details
p.8 #4Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
bjornthun wrote:
«Ai-S with ROM chip», is this a Dandelion chip?



I was speaking about the "CPU" on the SL-IIs line and the ZF.2/Milvus for Nikon. Is that what you're calling a dandelion chip?
Oct 16, 2021 at 08:22 PM
freaklikeme
Online
Details
p.8 #5Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
Fred Miranda wrote:
What I find interesting is that Sony users don't get the baked "distortion" correction in LR but Leica users do.
So L-mount shooters never get to see the uncorrected version unless they use different post-processing software apps like DxO PureRAW, RAWDigger, DNG Cleaner, etc..


The cynic in me says that was Leica's plan to keep people happy with their zooms and the Q. A "show people the beautiful sausage, not how it's made" approach.
Oct 16, 2021 at 08:28 PM
genji
Offline
Details
p.8 #6Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
freaklikeme wrote:
The cynic in me says that was Leica's plan to keep people happy with their zooms and the Q. A "show people the beautiful sausage, not how it's made" approach.


Not the slightest bit cynical, IMO. The uncorrected barrel distortion of the Q’s “Summilux” 28/1.7 has to be seen to be believed.
Oct 16, 2021 at 09:02 PM
TakenWild
Offline
Details
p.8 #7Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
DXO’s pure raw can remove baked in corrections from RAW files. DJI drones have an enormous amount of distortion correction baked into the RAW files as well. Probably how they keep the lenses so small. They are far worse than the sigmas.

Fred Miranda wrote:
DavidBM wrote
Fred Miranda wrote
DavidBM wrote
Fred Miranda wrote
DavidBM wroteSo it must have the non discretionary manufacturer profile turned on for distortion. More usual for just LaCA. In a way that’s a good sign….that sharpness across the field is after correction! Promising!

Some Sony lenses for APS have had non-discretionary manufacturer profiles in LR for distortion if I recall….maybe also the 2/28? But I’m not sure…
This is the first time I see distortion included in a LR built-in profile. With Sony full frame lenses, there is only automated correction for "lateral CA" and that includes the Sony 28mm f/2.

Yes, both 24mm and 90mm DG DN are sharp to the edges and improve slightly stopping down. They are not the sharpest I've seen but perform at very high level. The 23-year-old Leica APO-Summicron-M 90 f/2 ASPH @f/2.8 beats it with only 5 elements. However, when comparing both lenses wide open (f/2 vs f/2.8), they are very similar at infinity.

Also similar stopped down to At f/4-5.6 across the field. The Leica is still a tad better even at f/8 at pixel level but it would be hard to differentiate them in real world images. Great value for a well-built $650 compact lens with AF!

All this with distortion correction already applied for the Sigma.
Ah, sorry, must have misremembered about compulsory distortion correction on the 2/28! It needed it!
This is OT for this thread but I'm shocked by the level of distortion from the new 24/2 DG DN and 90/2.8 lenses.
However, this could be a blessing in disguise. Read on...

Whenever the correction is "barrel", it requires image cropping and in the case of the Sigma 24/2 and Sony 28/2 lenses, it requires a lot of cropping. The difference between the two is that the Sigma is uber sharp before "distortion correction + crop" and the Sony is not.

So, what Sigma did was to basically design a 22mm FL lens and let distortion go crazy relying on auto-correction in-camera to remove the barrel distortion and match the 24mm FL. However, in the process, a sharp and usable section of the image is set to be cropped to allow for the big barrel correction but some images do not require distortion correction and therefore could benefit from that extra range. It's almost like having a 21mm lens with high distortion and a 24mm lens with no distortion in one. The former is actually sharper at the very corners!

This is all hidden and done behind the scenes for the Leica SL2 and I have not checked the Sony yet. On the Leica, there is no way to turn distortion correction off in-camera or in-post (with LR or CaptureOne). So, to gather this information, I used the "DNG Cleaner" app which strips "cooked" lens correction + crop from DNG files, which in turn, allows me to see the real uncorrected RAW image.
Exactly. Allowing distortion allows the lens to be incredibly sharp uncorrected, and possibly therefore sharper even resized and *corrected* that you could have done for the same price and size if designed without distortion. As you say, the 2/28 was not incredibly sharp even before correction, and so after correction…

PS is it a crop on the Leica DNG or a crop and resize? Ie are the pixel dimensions of the corrected file the same as usual for the Leica files?
Here is an image showing why there is an actual crop with resize after the massive "barrel" distortion correction for the Sigma 24/2 DG DN. Look at the "white area" of the image (second set below). That area must be cropped out of the image after correction which in turn means less pixels afterwards.

So the image goes from 47.3 MP to 46.7 MP after correction.

BTW: This correction is at +29 (!!) in Lightroom. For the Sigma 90/2.8, its pincushion is not as severe at around: -13 which still way higher than average. The good news for the 90/2.8
...Show more →
Oct 16, 2021 at 09:29 PM
Fred Miranda
Offline
Admin
Upload & Sell: On
Details
p.8 #8Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
freaklikeme wrote:
Fred Miranda wroteWhat I find interesting is that Sony users don't get the baked "distortion" correction in LR but Leica users do.
So L-mount shooters never get to see the uncorrected version unless they use different post-processing software apps like DxO PureRAW, RAWDigger, DNG Cleaner, etc..
The cynic in me says that was Leica's plan to keep people happy with their zooms and the Q. A "show people the beautiful sausage, not how it's made" approach.


The SL APO primes I've tested to date (35, 50, 90) have negligible distortion. I haven't tried the SL 28/2 APO yet though.
The SL 24-90 zoom has baked distortion correction and at 24mm and 90mm it's pretty extreme. Actually at 24mm mechanical vignetting is visible before correction. Clearly a lens designed for software correction.
Leica users are not given the choice to see this. Perhaps it's better not to know.
Oct 16, 2021 at 09:33 PM
Fred Miranda
Offline
Admin
Upload & Sell: On
Details
p.8 #9Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
TakenWild wrote:
DXO’s pure raw can remove baked in corrections from RAW files. DJI drones have an enormous amount of distortion correction baked into the RAW files as well. Probably how they keep the lenses so small. They are far worse than the sigmas.

Fred Miranda wrote
DavidBM wrote
Fred Miranda wrote
DavidBM wrote
Fred Miranda wroteThis is the first time I see distortion included in a LR built-in profile. With Sony full frame lenses, there is only automated correction for "lateral CA" and that includes the Sony 28mm f/2.

Yes, both 24mm and 90mm DG DN are sharp to the edges and improve slightly stopping down. They are not the sharpest I've seen but perform at very high level. The 23-year-old Leica APO-Summicron-M 90 f/2 ASPH @f/2.8 beats it with only 5 elements. However, when comparing both lenses wide open (f/2 vs f/2.8), they are very similar at infinity.

Also similar stopped down to At f/4-5.6 across the field. The Leica is still a tad better even at f/8 at pixel level but it would be hard to differentiate them in real world images. Great value for a well-built $650 compact lens with AF!

All this with distortion correction already applied for the Sigma.
Ah, sorry, must have misremembered about compulsory distortion correction on the 2/28! It needed it!
This is OT for this thread but I'm shocked by the level of distortion from the new 24/2 DG DN and 90/2.8 lenses.
However, this could be a blessing in disguise. Read on...

Whenever the correction is "barrel", it requires image cropping and in the case of the Sigma 24/2 and Sony 28/2 lenses, it requires a lot of cropping. The difference between the two is that the Sigma is uber sharp before "distortion correction + crop" and the Sony is not.

So, what Sigma did was to basically design a 22mm FL lens and let distortion go crazy relying on auto-correction in-camera to remove the barrel distortion and match the 24mm FL. However, in the process, a sharp and usable section of the image is set to be cropped to allow for the big barrel correction but some images do not require distortion correction and therefore could benefit from that extra range. It's almost like having a 21mm lens with high distortion and a 24mm lens with no distortion in one. The former is actually sharper at the very corners!

This is all hidden and done behind the scenes for the Leica SL2 and I have not checked the Sony yet. On the Leica, there is no way to turn distortion correction off in-camera or in-post (with LR or CaptureOne). So, to gather this information, I used the "DNG Cleaner" app which strips "cooked" lens correction + crop from DNG files, which in turn, allows me to see the real uncorrected RAW image.
Exactly. Allowing distortion allows the lens to be incredibly sharp uncorrected, and possibly therefore sharper even resized and *corrected* that you could have done for the same price and size if designed without distortion. As you say, the 2/28 was not incredibly sharp even before correction, and so after correction…

PS is it a crop on the Leica DNG or a crop and resize? Ie are the pixel dimensions of the corrected file the same as usual for the Leica files?
Here is an image showing why there is an actual crop with resize after the massive "barrel" distortion correction for the Sigma 24/2 DG DN. Look at the "white area" of the image (second set below). That area must be cropped out of the image after correction which in turn means less pixels afterwards.

So the image goes from 47.3 MP to 46.7 MP after correction.

BTW: This correction is at +33 (!!) in Lightroom. For the Sigma 90/2.8, its pincushion is not as severe at around: -13 which still way higher than average. The good news for the 90/2.8 is that since there is minimal crop and the correction is not as massive, loss of high frequency detail is minimal at the very corners. (I checked)

What I find interesting is that Sony users don't get the baked "distortion" correction in LR but Leica users do.
So L-mount shooters never get to see the uncorrected version unless they use different post-processing software apps like DxO PureRAW, RAWDigger, DNG Cleaner, etc..


Yes it does. It's the first app I tried. The problem is that it also applies noise reduction so I looked for alternatives. I recommend DNG Cleaner. It's about $10 and can run batches (Mac only though).
Iridient Developer and C1 also offer the choice to have lens profile "on" or "off". I wish Adobe would do the same.
Oct 16, 2021 at 09:38 PM
philip_pj
Offline
Details
p.8 #10Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
Data sheets are excellent examples of manufacturer correction and reporting standards. They are intended mostly for those with some technical interest in lens performance and characteristics. From recent to oldest in Zeiss's 21C lenses:

Each and every Batis lens shows only distortion turned 'on', even the 135mm which really should not need it. None report the uncorrected distortion, but inexplicably they report the focal distance. Sin of omission.

The Loxias either show actual distortion only (50mm, 85mm) or show both the uncorrected and corrected distortion curves (21mm, 25mm, 35mm). Honesty.

ALL of Zeiss's (mostly older) 'serious' lenses of the digital era show only actual distortion curves, no matter how painful - Milvus, Otus, ZEF, ZM. No 'value added', nothing to report.

Anyway: why do you want to know, customer of ours?
Oct 16, 2021 at 11:06 PM
freaklikeme
Online
Details
p.8 #11Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
Fred Miranda wrote:
freaklikeme wrote
Fred Miranda wroteWhat I find interesting is that Sony users don't get the baked "distortion" correction in LR but Leica users do.
So L-mount shooters never get to see the uncorrected version unless they use different post-processing software apps like DxO PureRAW, RAWDigger, DNG Cleaner, etc..
The cynic in me says that was Leica's plan to keep people happy with their zooms and the Q. A "show people the beautiful sausage, not how it's made" approach.
The SL APO primes I've tested to date (35, 50, 90) have negligible distortion. I haven't tried the SL 28/2 APO yet though.
The SL 24-90 zoom has baked distortion correction and at 24mm and 90mm it's pretty extreme. Actually at 24mm mechanical vignetting is visible before correction. Clearly a lens designed for software correction.
Leica users are not given the choice to see this. Perhaps it's better not to know.


I think that's why the chief complaint of the 24-90 is it's size. I also think that people would've had fewer complaints about the Sony 24-70/4 if they'd implemented a similar strategy (and been more consistent in production).
Oct 16, 2021 at 11:11 PM
fjablo
Online
Details
p.8 #12Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
Kind of OT but optical correction of distortion is also not lossless, but leads to other compromises in the optical design. Of course fine if some prefer optically corrected lenses, but digital correction is now also a valid tool in lens design. And the Leica Q is maybe the best example of that

It will be so interesting to see the comparison between the Skopar and the Sigma 90mm: one had to correct distortion optically, the other had to be capable of AF, both targeting roughly the same size & price point.
Oct 17, 2021 at 12:02 AM
bjornthun
Offline
Details
p.8 #13Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
freaklikeme wrote:
bjornthun wrote«Ai-S with ROM chip», is this a Dandelion chip?
I was speaking about the "CPU" on the SL-IIs line and the ZF.2/Milvus for Nikon. Is that what you're calling a dandelion chip?


No, not at all. The factory installed COUs in modern Voigtländer SL-IIs line and Zeiss ZF.2 lenses are the real deal. Dandelion chips were crappy, could disintegrate by themselves and wre bought off of eBay. You would glue those to the lens and program them from the camera.
Oct 17, 2021 at 02:11 AM
bjornthun
Offline
Details
p.8 #14Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
Fred Miranda wrote:
DavidBM wrote
Fred Miranda wrote
DavidBM wrote
Fred Miranda wrote
DavidBM wroteSo it must have the non discretionary manufacturer profile turned on for distortion. More usual for just LaCA. In a way that’s a good sign….that sharpness across the field is after correction! Promising!

Some Sony lenses for APS have had non-discretionary manufacturer profiles in LR for distortion if I recall….maybe also the 2/28? But I’m not sure…
This is the first time I see distortion included in a LR built-in profile. With Sony full frame lenses, there is only automated correction for "lateral CA" and that includes the Sony 28mm f/2.

Yes, both 24mm and 90mm DG DN are sharp to the edges and improve slightly stopping down. They are not the sharpest I've seen but perform at very high level. The 23-year-old Leica APO-Summicron-M 90 f/2 ASPH @f/2.8 beats it with only 5 elements. However, when comparing both lenses wide open (f/2 vs f/2.8), they are very similar at infinity.

Also similar stopped down to At f/4-5.6 across the field. The Leica is still a tad better even at f/8 at pixel level but it would be hard to differentiate them in real world images. Great value for a well-built $650 compact lens with AF!

All this with distortion correction already applied for the Sigma.
Ah, sorry, must have misremembered about compulsory distortion correction on the 2/28! It needed it!
This is OT for this thread but I'm shocked by the level of distortion from the new 24/2 DG DN and 90/2.8 lenses.
However, this could be a blessing in disguise. Read on...

Whenever the correction is "barrel", it requires image cropping and in the case of the Sigma 24/2 and Sony 28/2 lenses, it requires a lot of cropping. The difference between the two is that the Sigma is uber sharp before "distortion correction + crop" and the Sony is not.

So, what Sigma did was to basically design a 22mm FL lens and let distortion go crazy relying on auto-correction in-camera to remove the barrel distortion and match the 24mm FL. However, in the process, a sharp and usable section of the image is set to be cropped to allow for the big barrel correction but some images do not require distortion correction and therefore could benefit from that extra range. It's almost like having a 21mm lens with high distortion and a 24mm lens with no distortion in one. The former is actually sharper at the very corners!

This is all hidden and done behind the scenes for the Leica SL2 and I have not checked the Sony yet. On the Leica, there is no way to turn distortion correction off in-camera or in-post (with LR or CaptureOne). So, to gather this information, I used the "DNG Cleaner" app which strips "cooked" lens correction + crop from DNG files, which in turn, allows me to see the real uncorrected RAW image.
Exactly. Allowing distortion allows the lens to be incredibly sharp uncorrected, and possibly therefore sharper even resized and *corrected* that you could have done for the same price and size if designed without distortion. As you say, the 2/28 was not incredibly sharp even before correction, and so after correction…

PS is it a crop on the Leica DNG or a crop and resize? Ie are the pixel dimensions of the corrected file the same as usual for the Leica files?
Here is an image showing why there is an actual crop with resize after the massive "barrel" distortion correction for the Sigma 24/2 DG DN. Look at the "white area" of the image (second set below). That area must be cropped out of the image after correction which in turn means less pixels afterwards.

So the image goes from 47.3 MP to 46.7 MP after correction.

BTW: This correction is at +29 (!!) in Lightroom. For the Sigma 90/2.8, its pincushion is not as severe at around: -13 which still way higher than average. The good news for the 90/2.8
...Show more →

From 47.3 MP to 46.7 MP means losing 0.6 MP. Is that all you lose through digital correction of distortion? That’s only 1.27% of the total and should be imperceptible. Or, am I missing something here?
Oct 17, 2021 at 02:15 AM
fjablo
Online
Details
p.8 #15Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
bjornthun wrote:
Fred Miranda wrote
DavidBM wrote
Fred Miranda wrote
DavidBM wrote
Fred Miranda wroteThis is the first time I see distortion included in a LR built-in profile. With Sony full frame lenses, there is only automated correction for "lateral CA" and that includes the Sony 28mm f/2.

Yes, both 24mm and 90mm DG DN are sharp to the edges and improve slightly stopping down. They are not the sharpest I've seen but perform at very high level. The 23-year-old Leica APO-Summicron-M 90 f/2 ASPH @f/2.8 beats it with only 5 elements. However, when comparing both lenses wide open (f/2 vs f/2.8), they are very similar at infinity.

Also similar stopped down to At f/4-5.6 across the field. The Leica is still a tad better even at f/8 at pixel level but it would be hard to differentiate them in real world images. Great value for a well-built $650 compact lens with AF!

All this with distortion correction already applied for the Sigma.
Ah, sorry, must have misremembered about compulsory distortion correction on the 2/28! It needed it!
This is OT for this thread but I'm shocked by the level of distortion from the new 24/2 DG DN and 90/2.8 lenses.
However, this could be a blessing in disguise. Read on...

Whenever the correction is "barrel", it requires image cropping and in the case of the Sigma 24/2 and Sony 28/2 lenses, it requires a lot of cropping. The difference between the two is that the Sigma is uber sharp before "distortion correction + crop" and the Sony is not.

So, what Sigma did was to basically design a 22mm FL lens and let distortion go crazy relying on auto-correction in-camera to remove the barrel distortion and match the 24mm FL. However, in the process, a sharp and usable section of the image is set to be cropped to allow for the big barrel correction but some images do not require distortion correction and therefore could benefit from that extra range. It's almost like having a 21mm lens with high distortion and a 24mm lens with no distortion in one. The former is actually sharper at the very corners!

This is all hidden and done behind the scenes for the Leica SL2 and I have not checked the Sony yet. On the Leica, there is no way to turn distortion correction off in-camera or in-post (with LR or CaptureOne). So, to gather this information, I used the "DNG Cleaner" app which strips "cooked" lens correction + crop from DNG files, which in turn, allows me to see the real uncorrected RAW image.
Exactly. Allowing distortion allows the lens to be incredibly sharp uncorrected, and possibly therefore sharper even resized and *corrected* that you could have done for the same price and size if designed without distortion. As you say, the 2/28 was not incredibly sharp even before correction, and so after correction…

PS is it a crop on the Leica DNG or a crop and resize? Ie are the pixel dimensions of the corrected file the same as usual for the Leica files?
Here is an image showing why there is an actual crop with resize after the massive "barrel" distortion correction for the Sigma 24/2 DG DN. Look at the "white area" of the image (second set below). That area must be cropped out of the image after correction which in turn means less pixels afterwards.

So the image goes from 47.3 MP to 46.7 MP after correction.

BTW: This correction is at +33 (!!) in Lightroom. For the Sigma 90/2.8, its pincushion is not as severe at around: -13 which still way higher than average. The good news for the 90/2.8 is that since there is minimal crop and the correction is not as massive, loss of high frequency detail is minimal at the very corners. (I checked)

What I find interesting is that Sony users don't get the baked "distortion" correction in LR but Leica users do.
So L-mount shooters never get to see the uncorrected version unless they use different post-processing software apps like DxO PureRAW, RAWDigger, Iridient, DNG Cleaner, etc..
From 47.3 MP to 46.7 MP means losing 0.6 MP. Is that all you lose through digital correction of distortion? That’s only 1.27% of the total and should be imperceptible. Or, am I missing something here?


The warping/ stretching of the pixels also costs some sharpness
Oct 17, 2021 at 02:40 AM
LBJ2
Offline
Details
p.8 #16Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
Fred Miranda wrote:
TakenWild wroteDXO’s pure raw can remove baked in corrections from RAW files. DJI drones have an enormous amount of distortion correction baked into the RAW files as well. Probably how they keep the lenses so small. They are far worse than the sigmas.

Fred Miranda wrote
DavidBM wrote
Fred Miranda wrote
DavidBM wroteAh, sorry, must have misremembered about compulsory distortion correction on the 2/28! It needed it!
This is OT for this thread but I'm shocked by the level of distortion from the new 24/2 DG DN and 90/2.8 lenses.
However, this could be a blessing in disguise. Read on...

Whenever the correction is "barrel", it requires image cropping and in the case of the Sigma 24/2 and Sony 28/2 lenses, it requires a lot of cropping. The difference between the two is that the Sigma is uber sharp before "distortion correction + crop" and the Sony is not.

So, what Sigma did was to basically design a 22mm FL lens and let distortion go crazy relying on auto-correction in-camera to remove the barrel distortion and match the 24mm FL. However, in the process, a sharp and usable section of the image is set to be cropped to allow for the big barrel correction but some images do not require distortion correction and therefore could benefit from that extra range. It's almost like having a 21mm lens with high distortion and a 24mm lens with no distortion in one. The former is actually sharper at the very corners!

This is all hidden and done behind the scenes for the Leica SL2 and I have not checked the Sony yet. On the Leica, there is no way to turn distortion correction off in-camera or in-post (with LR or CaptureOne). So, to gather this information, I used the "DNG Cleaner" app which strips "cooked" lens correction + crop from DNG files, which in turn, allows me to see the real uncorrected RAW image.
Exactly. Allowing distortion allows the lens to be incredibly sharp uncorrected, and possibly therefore sharper even resized and *corrected* that you could have done for the same price and size if designed without distortion. As you say, the 2/28 was not incredibly sharp even before correction, and so after correction…

PS is it a crop on the Leica DNG or a crop and resize? Ie are the pixel dimensions of the corrected file the same as usual for the Leica files?
Here is an image showing why there is an actual crop with resize after the massive "barrel" distortion correction for the Sigma 24/2 DG DN. Look at the "white area" of the image (second set below). That area must be cropped out of the image after correction which in turn means less pixels afterwards.

So the image goes from 47.3 MP to 46.7 MP after correction.

BTW: This correction is at +33 (!!) in Lightroom. For the Sigma 90/2.8, its pincushion is not as severe at around: -13 which still way higher than average. The good news for the 90/2.8 is that since there is minimal crop and the correction is not as massive, loss of high frequency detail is minimal at the very corners. (I checked)

What I find interesting is that Sony users don't get the baked "distortion" correction in LR but Leica users do.
So L-mount shooters never get to see the uncorrected version unless they use different post-processing software apps like DxO PureRAW, RAWDigger, DNG Cleaner, etc..
Yes it does. It's the first app I tried. The problem is that it also applies noise reduction so I looked for alternatives. I recommend DNG Cleaner. It's about $10 and can run batches (Mac only though).
Iridient Developer gives you a choice to have lens profile "on" or "off". I wish Adobe would do the same.


I've already run a few typical compositions SL2 SL 35/50/90 DNG files through the latest version of the base Raw Digger app and compared to the same DNG files in LR. But I'm not sure if Raw Digger presents lens distortions other than vignetting since I really didn't see distortions in these SL2 prime files in Raw Digger? I never used Raw Digger for this purpose before. *Or maybe I need to take a photo sample of a flat wall like you did in your earlier examples of this thread to better scrutinize?

Either way, I am also going to try the DNG Cleaner app to compare with what I saw in Raw Digger.
Oct 17, 2021 at 08:05 AM
Fred Miranda
Offline
Admin
Upload & Sell: On
Details
p.8 #17Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
bjornthun wrote:
Fred Miranda wrote
DavidBM wrote
Fred Miranda wrote
DavidBM wrote
Fred Miranda wroteThis is the first time I see distortion included in a LR built-in profile. With Sony full frame lenses, there is only automated correction for "lateral CA" and that includes the Sony 28mm f/2.

Yes, both 24mm and 90mm DG DN are sharp to the edges and improve slightly stopping down. They are not the sharpest I've seen but perform at very high level. The 23-year-old Leica APO-Summicron-M 90 f/2 ASPH @f/2.8 beats it with only 5 elements. However, when comparing both lenses wide open (f/2 vs f/2.8), they are very similar at infinity.

Also similar stopped down to At f/4-5.6 across the field. The Leica is still a tad better even at f/8 at pixel level but it would be hard to differentiate them in real world images. Great value for a well-built $650 compact lens with AF!

All this with distortion correction already applied for the Sigma.
Ah, sorry, must have misremembered about compulsory distortion correction on the 2/28! It needed it!
This is OT for this thread but I'm shocked by the level of distortion from the new 24/2 DG DN and 90/2.8 lenses.
However, this could be a blessing in disguise. Read on...

Whenever the correction is "barrel", it requires image cropping and in the case of the Sigma 24/2 and Sony 28/2 lenses, it requires a lot of cropping. The difference between the two is that the Sigma is uber sharp before "distortion correction + crop" and the Sony is not.

So, what Sigma did was to basically design a 22mm FL lens and let distortion go crazy relying on auto-correction in-camera to remove the barrel distortion and match the 24mm FL. However, in the process, a sharp and usable section of the image is set to be cropped to allow for the big barrel correction but some images do not require distortion correction and therefore could benefit from that extra range. It's almost like having a 21mm lens with high distortion and a 24mm lens with no distortion in one. The former is actually sharper at the very corners!

This is all hidden and done behind the scenes for the Leica SL2 and I have not checked the Sony yet. On the Leica, there is no way to turn distortion correction off in-camera or in-post (with LR or CaptureOne). So, to gather this information, I used the "DNG Cleaner" app which strips "cooked" lens correction + crop from DNG files, which in turn, allows me to see the real uncorrected RAW image.
Exactly. Allowing distortion allows the lens to be incredibly sharp uncorrected, and possibly therefore sharper even resized and *corrected* that you could have done for the same price and size if designed without distortion. As you say, the 2/28 was not incredibly sharp even before correction, and so after correction…

PS is it a crop on the Leica DNG or a crop and resize? Ie are the pixel dimensions of the corrected file the same as usual for the Leica files?
Here is an image showing why there is an actual crop with resize after the massive "barrel" distortion correction for the Sigma 24/2 DG DN. Look at the "white area" of the image (second set below). That area must be cropped out of the image after correction which in turn means less pixels afterwards.

So the image goes from 47.3 MP to 46.7 MP after correction.

BTW: This correction is at +33 (!!) in Lightroom. For the Sigma 90/2.8, its pincushion is not as severe at around: -13 which still way higher than average. The good news for the 90/2.8 is that since there is minimal crop and the correction is not as massive, loss of high frequency detail is minimal at the very corners. (I checked)

What I find interesting is that Sony users don't get the baked "distortion" correction in LR but Leica users do.
So L-mount shooters never get to see the uncorrected version unless they use different post-processing software apps like DxO PureRAW, RAWDigger, Iridient, DNG Cleaner, etc..
From 47.3 MP to 46.7 MP means losing 0.6 MP. Is that all you lose through digital correction of distortion? That’s only 1.27% of the total and should be imperceptible. Or, am I missing something here?


---------------------------------------------

fjablo wrote:
bjornthun wrote
Fred Miranda wrote
DavidBM wrote
Fred Miranda wrote
DavidBM wroteAh, sorry, must have misremembered about compulsory distortion correction on the 2/28! It needed it!
This is OT for this thread but I'm shocked by the level of distortion from the new 24/2 DG DN and 90/2.8 lenses.
However, this could be a blessing in disguise. Read on...

Whenever the correction is "barrel", it requires image cropping and in the case of the Sigma 24/2 and Sony 28/2 lenses, it requires a lot of cropping. The difference between the two is that the Sigma is uber sharp before "distortion correction + crop" and the Sony is not.

So, what Sigma did was to basically design a 22mm FL lens and let distortion go crazy relying on auto-correction in-camera to remove the barrel distortion and match the 24mm FL. However, in the process, a sharp and usable section of the image is set to be cropped to allow for the big barrel correction but some images do not require distortion correction and therefore could benefit from that extra range. It's almost like having a 21mm lens with high distortion and a 24mm lens with no distortion in one. The former is actually sharper at the very corners!

This is all hidden and done behind the scenes for the Leica SL2 and I have not checked the Sony yet. On the Leica, there is no way to turn distortion correction off in-camera or in-post (with LR or CaptureOne). So, to gather this information, I used the "DNG Cleaner" app which strips "cooked" lens correction + crop from DNG files, which in turn, allows me to see the real uncorrected RAW image.
Exactly. Allowing distortion allows the lens to be incredibly sharp uncorrected, and possibly therefore sharper even resized and *corrected* that you could have done for the same price and size if designed without distortion. As you say, the 2/28 was not incredibly sharp even before correction, and so after correction…

PS is it a crop on the Leica DNG or a crop and resize? Ie are the pixel dimensions of the corrected file the same as usual for the Leica files?
Here is an image showing why there is an actual crop with resize after the massive "barrel" distortion correction for the Sigma 24/2 DG DN. Look at the "white area" of the image (second set below). That area must be cropped out of the image after correction which in turn means less pixels afterwards.

So the image goes from 47.3 MP to 46.7 MP after correction.

BTW: This correction is at +33 (!!) in Lightroom. For the Sigma 90/2.8, its pincushion is not as severe at around: -13 which still way higher than average. The good news for the 90/2.8 is that since there is minimal crop and the correction is not as massive, loss of high frequency detail is minimal at the very corners. (I checked)

What I find interesting is that Sony users don't get the baked "distortion" correction in LR but Leica users do.
So L-mount shooters never get to see the uncorrected version unless they use different post-processing software apps like DxO PureRAW, RAWDigger, Iridient, DNG Cleaner, etc..
From 47.3 MP to 46.7 MP means losing 0.6 MP. Is that all you lose through digital correction of distortion? That’s only 1.27% of the total and should be imperceptible. Or, am I missing something here?
The warping/ stretching of the pixels also costs some sharpness


That was the megapixel number displayed before (uncorrected DNG image) and after automatic and irreversible profile correction in LR.
However things are actually much worse than that when looking at the actual pixels before and after 'correction + constrain crop'. The real difference is more like 34%. Read on...

Files from the 24/2 DG DN "L-mount" show up in Lightroom as 46.7MP (which is the default SL2's MP sensor). Like I wrote, when editing in Lightroom, distortion correction is automatic and irreversible for these L-mount DNG files.

If I run the original DNG file through DNG Cleaner, I get a slightly bigger file at 47.3MP which is slightly larger than the default SL2 files.

Now, if I load the "uncorrected" 47.3MP DNG file in LR and match the distortion correction + constrain crop done automatically for the same regular 46.7MP DNG (auto-corrected version), the number of pixels are reduced drastically since the correction + crops is so big. It goes from 8392 x 5632 to 7177 x 4817, which means ~47MP to ~35MP. So, effectively there is a noticeable loss of resolution/contrast off-axis due to stretching and interpolation.

Personally I am not a fan of this optical design compromise but as @fjablo wrote, there are also compromises to IQ, size and price when correcting distortion optically.




Oct 17, 2021 at 11:59 AM
bjornthun
Offline
Details
p.8 #18Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
Fred Miranda wrote:
bjornthun wrote
Fred Miranda wrote
DavidBM wrote
Fred Miranda wrote
DavidBM wroteAh, sorry, must have misremembered about compulsory distortion correction on the 2/28! It needed it!
This is OT for this thread but I'm shocked by the level of distortion from the new 24/2 DG DN and 90/2.8 lenses.
However, this could be a blessing in disguise. Read on...

Whenever the correction is "barrel", it requires image cropping and in the case of the Sigma 24/2 and Sony 28/2 lenses, it requires a lot of cropping. The difference between the two is that the Sigma is uber sharp before "distortion correction + crop" and the Sony is not.

So, what Sigma did was to basically design a 22mm FL lens and let distortion go crazy relying on auto-correction in-camera to remove the barrel distortion and match the 24mm FL. However, in the process, a sharp and usable section of the image is set to be cropped to allow for the big barrel correction but some images do not require distortion correction and therefore could benefit from that extra range. It's almost like having a 21mm lens with high distortion and a 24mm lens with no distortion in one. The former is actually sharper at the very corners!

This is all hidden and done behind the scenes for the Leica SL2 and I have not checked the Sony yet. On the Leica, there is no way to turn distortion correction off in-camera or in-post (with LR or CaptureOne). So, to gather this information, I used the "DNG Cleaner" app which strips "cooked" lens correction + crop from DNG files, which in turn, allows me to see the real uncorrected RAW image.
Exactly. Allowing distortion allows the lens to be incredibly sharp uncorrected, and possibly therefore sharper even resized and *corrected* that you could have done for the same price and size if designed without distortion. As you say, the 2/28 was not incredibly sharp even before correction, and so after correction…

PS is it a crop on the Leica DNG or a crop and resize? Ie are the pixel dimensions of the corrected file the same as usual for the Leica files?
Here is an image showing why there is an actual crop with resize after the massive "barrel" distortion correction for the Sigma 24/2 DG DN. Look at the "white area" of the image (second set below). That area must be cropped out of the image after correction which in turn means less pixels afterwards.

So the image goes from 47.3 MP to 46.7 MP after correction.

BTW: This correction is at +33 (!!) in Lightroom. For the Sigma 90/2.8, its pincushion is not as severe at around: -13 which still way higher than average. The good news for the 90/2.8 is that since there is minimal crop and the correction is not as massive, loss of high frequency detail is minimal at the very corners. (I checked)

What I find interesting is that Sony users don't get the baked "distortion" correction in LR but Leica users do.
So L-mount shooters never get to see the uncorrected version unless they use different post-processing software apps like DxO PureRAW, RAWDigger, Iridient, DNG Cleaner, etc..
From 47.3 MP to 46.7 MP means losing 0.6 MP. Is that all you lose through digital correction of distortion? That’s only 1.27% of the total and should be imperceptible. Or, am I missing something here?
---------------------------------------------

fjablo wrote
bjornthun wrote
Fred Miranda wrote
DavidBM wrote
Fred Miranda wroteThis is OT for this thread but I'm shocked by the level of distortion from the new 24/2 DG DN and 90/2.8 lenses.
However, this could be a blessing in disguise. Read on...

Whenever the correction is "barrel", it requires image cropping and in the case of the Sigma 24/2 and Sony 28/2 lenses, it requires a lot of cropping. The difference between the two is that the Sigma is uber sharp before "distortion correction + crop" and the Sony is not.

So, what Sigma did was to basically design a 22mm FL lens and let distortion go crazy relying on auto-correction in-camera to remove the barrel distortion and match the 24mm FL. However, in the process, a sharp and usable section of the image is set to be cropped to allow for the big barrel correction but some images do not require distortion correction and therefore could benefit from that extra range. It's almost like having a 21mm lens with high distortion and a 24mm lens with no distortion in one. The former is actually sharper at the very corners!

This is all hidden and done behind the scenes for the Leica SL2 and I have not checked the Sony yet. On the Leica, there is no way to turn distortion correction off in-camera or in-post (with LR or CaptureOne). So, to gather this information, I used the "DNG Cleaner" app which strips "cooked" lens correction + crop from DNG files, which in turn, allows me to see the real uncorrected RAW image.
Exactly. Allowing distortion allows the lens to be incredibly sharp uncorrected, and possibly therefore sharper even resized and *corrected* that you could have done for the same price and size if designed without distortion. As you say, the 2/28 was not incredibly sharp even before correction, and so after correction…

PS is it a crop on the Leica DNG or a crop and resize? Ie are the pixel dimensions of the corrected file the same as usual for the Leica files?
Here is an image showing why there is an actual crop with resize after the massive "barrel" distortion correction for the Sigma 24/2 DG DN. Look at the "white area" of the image (second set below). That area must be cropped out of the image after correction which in turn means less pixels afterwards.

So the image goes from 47.3 MP to 46.7 MP after correction.

BTW: This correction is at +33 (!!) in Lightroom. For the Sigma 90/2.8, its pincushion is not as severe at around: -13 which still way higher than average. The good news for the 90/2.8 is that since there is minimal crop and the correction is not as massive, loss of high frequency detail is minimal at the very corners. (I checked)

What I find interesting is that Sony users don't get the baked "distortion" correction in LR but Leica users do.
So L-mount shooters never get to see the uncorrected version unless they use different post-processing software apps like DxO PureRAW, RAWDigger, Iridient, DNG Cleaner, etc..
From 47.3 MP to 46.7 MP means losing 0.6 MP. Is that all you lose through digital correction of distortion? That’s only 1.27% of the total and should be imperceptible. Or, am I missing something here?
The warping/ stretching of the pixels also costs some sharpness
That was the megapixel number displayed before (uncorrected DNG image) and after automatic and irreversible profile correction in LR.
However things are actually much worse than that when looking at the actual pixels before and after 'correction + constrain crop'. The real difference is more like 34%. Read on...

Files from the 24/2 DG DN "L-mount" show up in Lightroom as 46.7MP (which is the default SL2's MP sensor). Like I wrote, when editing in Lightroom, distortion correction is automatic and irreversible for these L-mount DNG files.

If I run the original DNG file through DNG Cleaner, I get a slightly
...Show more →

Fred, how does the automatic profile do it. Does it first correct to distortion the same way you do manually and then crop, or does it do it differently. We could imagine a mathematical LR applying a mathematical formula (vector transformation), that essentially pulls at the corners of the image. The horisontal line cutting the image in to equal halves and the vertical line cutting the image in to equal halves would be unnaffected by such a transformation. The centre of the image would also be minimally impacted. The corner/peripheral parts would be the most affected. I imagine LR could do it in one go like this. This would impact far fewer pixels.

Do we know LR, C1 et al do it? Do each have its own method. LR in particular has its’ own profiles for many lenses. Does LR apply its own or the manufacturer’s profile, when profile correction is mandated by the manufacturer.?
Oct 17, 2021 at 12:16 PM
Fred Miranda
Offline
Admin
Upload & Sell: On
Details
p.8 #19Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
bjornthun wrote:
Fred Miranda wrote
bjornthun wrote
Fred Miranda wrote
DavidBM wrote
Fred Miranda wroteThis is OT for this thread but I'm shocked by the level of distortion from the new 24/2 DG DN and 90/2.8 lenses.
However, this could be a blessing in disguise. Read on...

Whenever the correction is "barrel", it requires image cropping and in the case of the Sigma 24/2 and Sony 28/2 lenses, it requires a lot of cropping. The difference between the two is that the Sigma is uber sharp before "distortion correction + crop" and the Sony is not.

So, what Sigma did was to basically design a 22mm FL lens and let distortion go crazy relying on auto-correction in-camera to remove the barrel distortion and match the 24mm FL. However, in the process, a sharp and usable section of the image is set to be cropped to allow for the big barrel correction but some images do not require distortion correction and therefore could benefit from that extra range. It's almost like having a 21mm lens with high distortion and a 24mm lens with no distortion in one. The former is actually sharper at the very corners!

This is all hidden and done behind the scenes for the Leica SL2 and I have not checked the Sony yet. On the Leica, there is no way to turn distortion correction off in-camera or in-post (with LR or CaptureOne). So, to gather this information, I used the "DNG Cleaner" app which strips "cooked" lens correction + crop from DNG files, which in turn, allows me to see the real uncorrected RAW image.
Exactly. Allowing distortion allows the lens to be incredibly sharp uncorrected, and possibly therefore sharper even resized and *corrected* that you could have done for the same price and size if designed without distortion. As you say, the 2/28 was not incredibly sharp even before correction, and so after correction…

PS is it a crop on the Leica DNG or a crop and resize? Ie are the pixel dimensions of the corrected file the same as usual for the Leica files?
Here is an image showing why there is an actual crop with resize after the massive "barrel" distortion correction for the Sigma 24/2 DG DN. Look at the "white area" of the image (second set below). That area must be cropped out of the image after correction which in turn means less pixels afterwards.

So the image goes from 47.3 MP to 46.7 MP after correction.

BTW: This correction is at +33 (!!) in Lightroom. For the Sigma 90/2.8, its pincushion is not as severe at around: -13 which still way higher than average. The good news for the 90/2.8 is that since there is minimal crop and the correction is not as massive, loss of high frequency detail is minimal at the very corners. (I checked)

What I find interesting is that Sony users don't get the baked "distortion" correction in LR but Leica users do.
So L-mount shooters never get to see the uncorrected version unless they use different post-processing software apps like DxO PureRAW, RAWDigger, Iridient, DNG Cleaner, etc..
From 47.3 MP to 46.7 MP means losing 0.6 MP. Is that all you lose through digital correction of distortion? That’s only 1.27% of the total and should be imperceptible. Or, am I missing something here?
---------------------------------------------

fjablo wrote
bjornthun wrote
Fred Miranda wrote
DavidBM wroteExactly. Allowing distortion allows the lens to be incredibly sharp uncorrected, and possibly therefore sharper even resized and *corrected* that you could have done for the same price and size if designed without distortion. As you say, the 2/28 was not incredibly sharp even before correction, and so after correction…

PS is it a crop on the Leica DNG or a crop and resize? Ie are the pixel dimensions of the corrected file the same as usual for the Leica files?
Here is an image showing why there is an actual crop with resize after the massive "barrel" distortion correction for the Sigma 24/2 DG DN. Look at the "white area" of the image (second set below). That area must be cropped out of the image after correction which in turn means less pixels afterwards.

So the image goes from 47.3 MP to 46.7 MP after correction.

BTW: This correction is at +33 (!!) in Lightroom. For the Sigma 90/2.8, its pincushion is not as severe at around: -13 which still way higher than average. The good news for the 90/2.8 is that since there is minimal crop and the correction is not as massive, loss of high frequency detail is minimal at the very corners. (I checked)

What I find interesting is that Sony users don't get the baked "distortion" correction in LR but Leica users do.
So L-mount shooters never get to see the uncorrected version unless they use different post-processing software apps like DxO PureRAW, RAWDigger, Iridient, DNG Cleaner, etc..
From 47.3 MP to 46.7 MP means losing 0.6 MP. Is that all you lose through digital correction of distortion? That’s only 1.27% of the total and should be imperceptible. Or, am I missing something here?
The warping/ stretching of the pixels also costs some sharpness
That was the megapixel number displayed before (uncorrected DNG image) and after automatic and irreversible profile correction in LR.
However things are actually much worse than that when looking at the actual pixels before and after 'correction + constrain crop'. The real difference is more like 34%. Read on...

Files from the 24/2 DG DN "L-mount" show up in Lightroom as 46.7MP (which is the default SL2's MP sensor). Like I wrote, when editing in Lightroom, distortion correction is automatic and irreversible for these L-mount DNG files.

If I run the original DNG file through DNG Cleaner, I get a slightly bigger file at 47.3MP which is slightly larger than the default SL2 files.

Now, if I load the "uncorrected" 47.3MP DNG file in LR and match the distortion correction + constrain crop done automatically for the same regular 46.7MP DNG (auto-corrected version), the number of pixels are reduced drastically since the correction + crops is so big. It goes from 8392 x 5632 to 7177 x 4817, which means ~47MP to ~35MP. So, effectively there is a noticeable loss of resolution/contrast off-axis due to stretching and interpolation.

Personally I am not a fan of this optical design compromise but as @fjablo@ wrote, there are also compromises to IQ, size and price when correcting distortion optically.
Fred, how does the automatic profile do it. Does it first correct to distortion the same way you do manually and then crop, or does it do it differently. We could imagine a mathematical LR applying a mathematical formula (vector transformation), that essentially pulls at the corners of the image. The horisontal line cutting the image in to equal halves and the vertical line cutting the image in to equal halves would be unnaffected by such a transformation. The centre of the image would also be minimally impacted. The corner/peripheral parts would be the most affected. I imagine LR could
...Show more →

Lots of questions and I wish I knew all the answers.

Here is what I know: Currently, Lightroom's "Built-in Lens Profile" is irreversible and automatically applied to supported lenses. I have just tried the latest version of C1 with L-mount DNG files and they are auto-corrected in the same fashion (LaCA + Distortion) by default. However, this can be reversed using the Lens Correction tab (Just like with Iridient Developer). So more power to the user and I wish Adobe would add that option as well.

I believe the software flags camera make/model since distortion correction is applied to Leica/Sigma/Panasonic DNG files but for some reason is not applied to Sony ARW files even when using the same Sigma DG DN lenses. This is unrelated to "in-camera" correction since the their "Built-in Lens Profile" takes place even if those settings are "off" in-camera.

You are correct that it does not affect the entire image but such huge barrel distortion correction affects unevenly most of the image frame. The worse ill effect from this type of correction will be more noticeable towards the edges where the stretching/interpolation is more severe.

Perhaps we will see new "AI" distortion correction soon, but right now there is noticeable IQ deterioration in the process.

The good news and trying to come back to something more related to the new Voigtlander 90/2.8 APO is that the distortion correction for the Sigma 90/2.8 is way less severe and IQ deterioration is minimal even when scrutinizing the edges at pixel level.
Oct 17, 2021 at 12:35 PM
Ulysseita
Offline
Details
p.8 #20Pre-order: Voigtlander APO-SKOPAR 90mm f/2.8 VM and F-mount
I confirm, Lightroom and CRaw receive the fix and keep it on the program, You can't remove them; just a reference, I had the lx100 by Panasonic, on Lightroom the correction was by default and impossible to change: on C1 I was able to use the whole frame on the wide end allowing me to start from 22mm instead of 24mm...

Oct 17, 2021 at 01:35 PM
FM ForumsLeica & Alternative GearJoin Upload & Sell
  
 

You are not logged inLog inRegister