Monito wrote:
Thanks. The results are as Dan Mitchell predicted:
After sharpening, the 5DS and 5DsR images are practically indistinguishable in detail.
But ... the 5DsR images clearly have rainbows of moiré.
I think I made the right decision then on not going forward with the 5Dr... but.. that isn't to say buying the "R" is a bad choice. Either camera IMO ... you can't go wrong. Still in awe over the images and the details everyone is posting.
gdanmitchell wrote:
I recommend a stop to the hyperventilating. Just take the camera and make photographs with it. You'll be fine. Really.
I have been using a PhaseOne 60MP IQ160 for a few years now. Moire has not really been a problem. I have shot fashion/portraits and landscapes. The detail and color (with capture one) has just been stunning. I like that the files do not require much sharpening most times.
I have seen some sporadic aliasing in some areas of some images but it has been easy to fix and only has happened when I used my best lenses at optimum apertures (f8 in the case of the Rodenstock HR lenses) If I stop down a bit or shoot wide open the slight softening takes care of most aliasing.
This was tested with the D800 / D800E. The E only had a resolution advantage in a very narrow aperture range.
It is good to have it available if you want it / need it.
snapsy wrote:
Here are the dpreview 5DS vs 5DR studio samples, orig vs my deconvolution sharpening in RawTherapee:
5DSR vs 5DS dpreview studio 100% crops
Thanks for the samples. Compression artifacts make it hard to analyze the difference. That aside, here is a cropped and magnified section:
The sharpened 5Ds is on the right. As expected, it has ringing in high frequency areas as I have highlighted.
I am sure we can make even lower resolution images look like 5DsR if we didn't care about artifacts and only shot for perceptual sharpness.
amirm wrote:
Thanks for the samples. Compression artifacts make it hard to analyze the difference. That aside, here is a cropped and magnified section:
The sharpened 5Ds is on the right. As expected, it has ringing in high frequency areas as I have highlighted.
I am sure we can make even lower resolution images look like 5DsR if we didn't care about artifacts and only shot for perceptual sharpness.
You're showing a 200% crop to exhibit sharpening artifacts? And naturally the "5DS sharpened" image has more sharpening than unsharpened 5DR. And the 5DS sharpening can easily be dialed down and still approach the sharpness of the 5DR. I only used RT's defaults for convolution.
A sample image. During the middle of the day in boring light I aimed the 5DsR up into a canyon not far from Tioga Pass. ISO 100. Live view with camera on the tripod. EF 70-200mm f/2.8L II at 120mm and f/5.6.
Full scene — look closely for crop marks around the small section shown below at 100% magnification.
Due to exposure issues with the original image I increased "exposure, contrast, shadows" in ACR, added my default "15" sharpening there, with the mask slider decreasing the effect away from edges. I did not apply my usual "clarity" setting or the sharpening that I would typically apply in Photoshop.
If you wanted to see how sharp a 30" print would be, simply keep the dimensions of the crop at 600 x 399 pixels while changing resolution to 300 dpi, do typical sharpening (I'd perhaps go with radius: .7 and amount: 150 or so) and make a very small print of this section — native 300 resolution gets you very close to 30" width from a full, uncorked 5Ds image.
Gunzorro wrote:
Yep! I bought a copy in near mint condition this year, along with the same vintage 35-105 f/3.5-4.5.
The 28-70 f/3.5-4.5 is a great little lens. It was originally the top pro lens for the then-new EF mount cameras, preceding the original 28-80 f/2.8-4.0L lens by two years for the first version, and one year on the version II. I have the version II. Mine was about $60, IIRC. Lots of fun, and very light weight. Decent optics and coatings. I got the lens hood too, and that helps the most to give good IQ.
Nice to see how it performs on the new 5DsR. Looking forward to more shots from you. ...Show more →
Wow you got $60 only .... the vintage 35-105 you mentioned my friend also tried on his 5DsR, come out result is also good (but I haven't seen the 100% crop yet), perhaps you can take some shots to see.
When I put the 28-70/ 3.5-4.5 on 5DsR, the camera can only show f/4.5 as minimum number, but I can tell from the picture I can use wide open f/3.5 @ 28mm. Perhaps the lens is too old to recognise by a new body.
gdanmitchell wrote:
A sample image. During the middle of the day in boring light I aimed the 5DsR up into a canyon not far from Tioga Pass. ISO 100. Live view with camera on the tripod. EF 70-200mm f/2.8L II at 120mm and f/5.6.
Full scene — look closely for crop marks around the small section shown below at 100% magnification.
Due to exposure issues with the original image I increased "exposure, contrast, shadows" in ACR, added my default "15" sharpening there, with the mask slider decreasing the effect away from edges. I did not apply my usual "clarity" setting or the sharpening that I would typically apply in Photoshop.
If you wanted to see how sharp a 30" print would be, simply keep the dimensions of the crop at 600 x 399 pixels while changing resolution to 300 dpi, do typical sharpening (I'd perhaps go with radius: .7 and amount: 150 or so) and make a very small print of this section — native 300 resolution gets you very close to 30" width from a full, uncorked 5Ds image.
I was pretty bummed out driving home from that shoot, thinking I hadn't captured anything worthwhile, but I was very pleasantly surprised when I started playing around with those files in LR the next morning.
I think it's just about time to dump the rest of my Sony stuff - before it depreciates to zero...
I just sold my A7s couple days ago, sold the whole system, to avoid the fast depreciation
gdanmitchell wrote:
A sample image. During the middle of the day in boring light I aimed the 5DsR up into a canyon not far from Tioga Pass. ISO 100. Live view with camera on the tripod. EF 70-200mm f/2.8L II at 120mm and f/5.6.
Full scene — look closely for crop marks around the small section shown below at 100% magnification.
Due to exposure issues with the original image I increased "exposure, contrast, shadows" in ACR, added my default "15" sharpening there, with the mask slider decreasing the effect away from edges. I did not apply my usual "clarity" setting or the sharpening that I would typically apply in Photoshop.
If you wanted to see how sharp a 30" print would be, simply keep the dimensions of the crop at 600 x 399 pixels while changing resolution to 300 dpi, do typical sharpening (I'd perhaps go with radius: .7 and amount: 150 or so) and make a very small print of this section — native 300 resolution gets you very close to 30" width from a full, uncorked 5Ds image.
Hi Dan, we need to share our sharpening recipes as they unfold. I initially though this camera would need less sharpening and so dropped my normal methods to about 60%. But I was playing yesterday and saw that these files can take a ton of sharpening without falling apart, showing halos and artifacts. Not that it needs it, but it seems worthwhile to experiment.
I plan to spend the day on this and might start a new thread. I also noted that the 5DS-R gets live view focus all the time at 1X. My 5D3 needed to be zoomed to 5X with the 11-24 at the wide end sometimes when the contrast was low or the light was low. An ultrawide makes stuff small. I suspect the extra pixels help the 5DS-R during focus but I don't know enough about how live view focus works to be sure.
ben egbert wrote:
Hi Dan, we need to share our sharpening recipes as they unfold. I initially though this camera would need less sharpening and so dropped my normal methods to about 60%. But I was playing yesterday and saw that these files can take a ton of sharpening without falling apart, showing halos and artifacts. Not that it needs it, but it seems worthwhile to experiment.
I plan to spend the day on this and might start a new thread. I also noted that the 5DS-R gets live view focus all the time at 1X. My 5D3 needed to be zoomed to 5X with the 11-24 at the wide end sometimes when the contrast was low or the light was low. An ultrawide makes stuff small. I suspect the extra pixels help the 5DS-R during focus but I don't know enough about how live view focus works to be sure....Show more →
About focus first. The 5Ds R can give 16x magnification for live view manual focus, so it seems that a great deal of manual focus precision is possible. Ultra-wides make this more challenging (with any live view camera) in two ways. As you point out, much of the small stuff that you might otherwise focus on is so small that it isn't so useful for that purpose — and you have to find other things to focus on. Secondly, because DOF is effectively larger on such lenses (due both to the shorter focal length and typically smaller maximum apertures) it is harder to clearly see when the image click in (and out) of ideal focus.
Regarding sharpening... I haven't yet worked with a sufficient number of files and I haven't printed anything from my 5Ds R yet, so I don't have a handle on this yet. My sharpening workflow usually has four components:
I do some initial mild sharpening in ACR, usually with amount set to about 15 and then some masking added to constrain the ACR sharpening to edges. The same settings seem to work here.
I also typically adjust the clarity slider in ACR, usually raising it slightly — a typical value might be about 12, though it can range from a negative value in some cases to a stronger positive value in others.
I sharpen for detail in Photoshop, too. Starting values are amount: 150 and radius: .7. My initial attempts with the 5DsR suggest that this won't have to change all that much.
Finally, I do output sharpening on the resized and flattened copy of the file that I will print. I've only tried this on one file, and I did not actually make the print. In the initial trial it looked to me like I may need to use a bit less print sharpening, but I won't know until I actually produce a print!
By the way, at every stage I do non-destructive sharpening — so that I can redo/undo any of these choices and try other values.
gdanmitchell wrote:
About focus first. The 5Ds R can give 16x magnification for live view manual focus, so it seems that a great deal of manual focus precision is possible. Ultra-wides make this more challenging (with any live view camera) in two ways. As you point out, much of the small stuff that you might otherwise focus on is so small that it isn't so useful for that purpose — and you have to find other things to focus on. Secondly, because DOF is effectively larger on such lenses (due both to the shorter focal length and typically smaller maximum apertures) it is harder to clearly see when the image click in (and out) of ideal focus.
Regarding sharpening... I haven't yet worked with a sufficient number of files and I haven't printed anything from my 5Ds R yet, so I don't have a handle on this yet. My sharpening workflow usually has four components:
I do some initial mild sharpening in ACR, usually with amount set to about 15 and then some masking added to constrain the ACR sharpening to edges. The same settings seem to work here.
I also typically adjust the clarity slider in ACR, usually raising it slightly — a typical value might be about 12, though it can range from a negative value in some cases to a stronger positive value in others.
I sharpen for detail in Photoshop, too. Starting values are amount: 150 and radius: .7. My initial attempts with the 5DsR suggest that this won't have to change all that much.
Finally, I do output sharpening on the resized and flattened copy of the file that I will print. I've only tried this on one file, and I did not actually make the print. In the initial trial it looked to me like I may need to use a bit less print sharpening, but I won't know until I actually produce a print!
By the way, at every stage I do non-destructive sharpening — so that I can redo/undo any of these choices and try other values.
Thanks Dan. Mine is similar in workflow but with different values. During ACR I have been using 50, 0.6, 7 with a starting mask of 30. I will be playing with these values today. I am not doing any clarity or large radius small amount now because I get this in the HDR program.
After HDR, I run two USM both at 300,0.3,0 faded to 35%. The first is lab, the second is RGB in darken mode. I will check out your values. I assume you are not fading.
For output, I let Qimage do the output sharpening for prints, and Topaz 333 for web after bicubic smoother downsize in one step. Topaz 333 is a pretty light sharpening. These are leftover from 5D3 so I need to revisit each of these.
Edit for Dan.
There are 2 more values in ACR sharpening, you give amount, 15, how about radius and detail?
Ming Thein posted his first thoughts about the 5DS-R here, with a full review to follow. I'm a big fan of Ming's. First, I think he's a fantastic photographer. Second, his reviews are entirely centered around real-world usage, with heavy focus on ergonomics, workflow, and IQ, particularly for high-quality very-large prints. He also has a history of finding issues that everyone else misses. For example he was the first to report the D800/E "Left AF" issues.
gdanmitchell wrote:
A sample image. During the middle of the day in boring light I aimed the 5DsR up into a canyon not far from Tioga Pass. ISO 100. Live view with camera on the tripod. EF 70-200mm f/2.8L II at 120mm and f/5.6.
Full scene — look closely for crop marks around the small section shown below at 100% magnification.
Due to exposure issues with the original image I increased "exposure, contrast, shadows" in ACR, added my default "15" sharpening there, with the mask slider decreasing the effect away from edges. I did not apply my usual "clarity" setting or the sharpening that I would typically apply in Photoshop.
If you wanted to see how sharp a 30" print would be, simply keep the dimensions of the crop at 600 x 399 pixels while changing resolution to 300 dpi, do typical sharpening (I'd perhaps go with radius: .7 and amount: 150 or so) and make a very small print of this section — native 300 resolution gets you very close to 30" width from a full, uncorked 5Ds image.
