This topic is in the 2015 archive. You can read it and like its posts here. It is read only.
Archive 2015No low pass filter
p.2 #1No low pass filter
Hmm, maybe I need to go back to film if I want to avoid sharpening in post.
Feb 04, 2015 at 07:20 PM
p.2 #2No low pass filter
Camperjim wrote:
The idea of an AA filter never made any sense to me. It always seemed like poor, stopgap engineering. Why blur everything just to avoid rare cases of moire? Using sharpening software does not restore what was lost and adds additional artifacts. One reason for high resolution sensors is to eliminate this poor design.
Aliasing is not so rare with sharp lenses in contrasty light. Fur and feathers, or even the human hair can be a real issue. I'm really not sure what to do either. I'd prefer a fairly weak AA filter, but am concerned that the 5Ds' filter will be strong to differentiate it from the 5DsR.
EBH
The idea of an AA filter never made any sense to me. It always seemed like poor, stopgap engineering. Why blur everything just to avoid rare cases of moire? Using sharpening software does not restore what was lost and adds additional artifacts. One reason for high resolution sensors is to eliminate this poor design.
Aliasing is not so rare with sharp lenses in contrasty light. Fur and feathers, or even the human hair can be a real issue. I'm really not sure what to do either. I'd prefer a fairly weak AA filter, but am concerned that the 5Ds' filter will be strong to differentiate it from the 5DsR.
EBH
Feb 04, 2015 at 07:24 PM
p.2 #3No low pass filter
ben egbert wrote:
Hmm, maybe I need to go back to film if I want to avoid sharpening in post.
So you never sharpened the scans? It's a normal part of the digital workflow to compensate for the scanning process.
Some films like Velvia did their own chemical sharpening, and of course you've probably seen some photo prints made with unsharp masks.
EBH
Hmm, maybe I need to go back to film if I want to avoid sharpening in post.
So you never sharpened the scans? It's a normal part of the digital workflow to compensate for the scanning process.
Some films like Velvia did their own chemical sharpening, and of course you've probably seen some photo prints made with unsharp masks.
EBH
Feb 04, 2015 at 07:28 PM
p.2 #4No low pass filter
I think practically speaking, no 35 mm film you ever used or are likely to ever use is as sharp as the 5D series of cameras, as an example, and certainly not the 5D3 and beyond.
If you are talking about 2 1/4 x 3 1/4 or 4x5 or 8x10, then yes, you could get higher resolution (hence sharper), but packing that kind of gear around for landscapes will make you realize you are not as young as you once were.
In any case, if your film or sensor out-resolves your lenses, you will want to sharpen in post. Sharp grain does not compensate for a slight bit of lens blur off-center.
The problem is not sharpening in post. For almost all work that can be effectively automated for capture sharpening and output sharpening. The creative sharpening in between is an area for advanced practitioners to get an extra edge (weak pun partially intended).
You can pine away for film but in the end I bet you don't go back except a little bit sometimes perhaps just for fun and the hands on.
If you are talking about 2 1/4 x 3 1/4 or 4x5 or 8x10, then yes, you could get higher resolution (hence sharper), but packing that kind of gear around for landscapes will make you realize you are not as young as you once were.
In any case, if your film or sensor out-resolves your lenses, you will want to sharpen in post. Sharp grain does not compensate for a slight bit of lens blur off-center.
The problem is not sharpening in post. For almost all work that can be effectively automated for capture sharpening and output sharpening. The creative sharpening in between is an area for advanced practitioners to get an extra edge (weak pun partially intended).
You can pine away for film but in the end I bet you don't go back except a little bit sometimes perhaps just for fun and the hands on.
Feb 04, 2015 at 07:32 PM
p.2 #5No low pass filter
Aliasing also shows up as false colours. In black hair, you can tone it down by applying varying amounts of monochrome filters to that area, but green tones in blond hair are more difficult to deal with.
Feb 04, 2015 at 07:34 PM
p.2 #6No low pass filter
EB-1 wrote:
So you never sharpened the scans? It's a normal part of the digital workflow to compensate for the scanning process.
Some films like Velvia did their own chemical sharpening, and of course you've probably seen some photo prints made with unsharp masks.
EBH
I never scanned or printed. Just project the slide when you get it back. Not that I am against pp, I just hate the fact that RAW's are so crippled out of the camera. I spend 30 seconds max per image sharpening so its only a psychological issue.
ben egbert wrote
Hmm, maybe I need to go back to film if I want to avoid sharpening in post.Some films like Velvia did their own chemical sharpening, and of course you've probably seen some photo prints made with unsharp masks.
EBH
I never scanned or printed. Just project the slide when you get it back. Not that I am against pp, I just hate the fact that RAW's are so crippled out of the camera. I spend 30 seconds max per image sharpening so its only a psychological issue.
Feb 04, 2015 at 07:41 PM
p.2 #7No low pass filter
I shoot almost exclusively with the Kodak SLR/C which has "NO" AA filter. The relationship between your lens sharpness and sensor and subject pattern will dictate to what degree your AA filter needs are.
If you go to maxmax.com @
http://maxmax.com/hot_rod_visible.htm
you'll see some discussion & images on the matter. Granted this includes some older images / cameras. But, the general gist is that the AA filter imparts an intentional blur as a safeguard to moire'.
The absence of the AA filter (a true non-AA) affords a sharper/more contrast image ... but at the potential peril of moire' / false coloration interpretation by the the Bayer sensor (longing for the non-Bayer array sensor sans AA filter).
For MY SHOOTING ... I have personally experienced moire' / false color ONCE (an image of a golfer @ his knit golf shirt, high contrast lighting). Because most of my shooting does not include subject matter that is typical of pattern frequency that is key to inducing such moire' ... I'm pretty safe @ shooting "naked" without the protection of an AA filter intentionally inducing a level of blur.
According to my discussions with MaxMax (a few years ago), the gains to be had by going without an AA filter are on the order of 15% - 20% resolution. That was of course relative to the time when we were discussing how a 12MP or 14MP non-AA could compare more closely to a 16MP-20MP camera @ resolving capability.
Fast forward to the prospect of a 50MP non-AA vs. a 50MP AA. Personally, I don't need 50MP ... but I do have a very strong preference toward a non-AA filter. My reasoning for this is that sharp / blur is an issue of contrast. By NOT having an AA involved, we are not reducing contrast @ capture, only to increase contrast in processing. The less contrast enhancement we impart, the less artifacts we risk inducing.
So ... non-AA incurs risk of moire' ... AA incurs risk of artifacts @ pushing contrast. As the pixels sizes become increasingly smaller, the issues of each become seemingly less detrimental. In order to incur moire' @ 50MP ... good grief, you are going to have to have some incredibly sharp glass and stellar technique to outresolve the sensor. OTOH, good grief, the diff @ 50MP AA vs. 50MP non-AA is going to be imperceptible to many as the contrast difference is mitigated by good PP and print sizes @ utility.
That being said ... as an ardent fan of non-AA ... unless you REALLY KNOW why you want an non-AA filter @ 50MP, then you probably don't really want a non-AA filter @ 50MP. At 14MP it can make quite a bit of diff. For most folks @ 50MP, they won't realize the diff. That said, I'd wait to see the comps from the cameras when they come out ... but in real world @ 50MP, FEW people will really warrant the diff @ non-AA vs. AA.
Of interest to note that some medium format digital cameras use removable AA filters, so they can utilize AA filter protection from moire' when shooting fashion, etc. and can remove the AA filter for non-pattern repeating subjects, such as landscape. Considering the limits of print, even an AA filter camera will likely outresolve paper on most days. So, on one hand you have your subject @ pattern vs. non-pattern. On the other hand you have lens resolving capability and print resolving capability ... all pieces of the puzzle regarding your decision for AA or non-AA. If you aren't that retentive about your shooting and processes involved ... skip the whole thing and just go with the AA filter like most folks.
Personally, I doubt my M645 glass will be sharp enough to outresolve a non-AA filter. Not sure about the now 100-400 5.6 II that is on my radar though. However, I won't be using a 100-400/5.6 II on fabric, so I expect to be safe. Conversely, if I do portraits with my M645 glass, I should be safe there also. IN one scenario, it is my subject matter that is preventing the moire' @ non-pattern repeating subject. In the other, it is the glass of choice inability to outresolve the sensor. Either way, moire' averted.
And, for those who shoot UWA, the resolving capability of UWA glass is typically much less than super long glass as the angles of incidence @ collimated light are significantly different to generate higher resolving capability in the longer glass. As a WA / UWA shooter ... outresolving a 50MP sensor is probably going to be a trick to achieve, even if you tried. For a birder, shooting long glass at a repeating feather pattern ... hmmmm @ non-AA for false color (but 50MP is really tiny).
For the OP ... if this has your head spinning ... borrow or rent a 50MP non-AA version and mount your sharpest lens. Shoot some fabric, some screens, some air conditioner grill vents, birds feathers, etc @ repeating patterns to try and induce some moire' / false color. Compare that to your shooting style and you can decide how the risk/reward tips your decision.
For me ... that's for me ... I'll suspect I'll always be a fan of non-AA (until I become a fashion photographer, anyway).
HTH
Edited on Feb 04, 2015 at 09:50 PM · View previous versions
If you go to maxmax.com @
http://maxmax.com/hot_rod_visible.htm
you'll see some discussion & images on the matter. Granted this includes some older images / cameras. But, the general gist is that the AA filter imparts an intentional blur as a safeguard to moire'.
The absence of the AA filter (a true non-AA) affords a sharper/more contrast image ... but at the potential peril of moire' / false coloration interpretation by the the Bayer sensor (longing for the non-Bayer array sensor sans AA filter).
For MY SHOOTING ... I have personally experienced moire' / false color ONCE (an image of a golfer @ his knit golf shirt, high contrast lighting). Because most of my shooting does not include subject matter that is typical of pattern frequency that is key to inducing such moire' ... I'm pretty safe @ shooting "naked" without the protection of an AA filter intentionally inducing a level of blur.
According to my discussions with MaxMax (a few years ago), the gains to be had by going without an AA filter are on the order of 15% - 20% resolution. That was of course relative to the time when we were discussing how a 12MP or 14MP non-AA could compare more closely to a 16MP-20MP camera @ resolving capability.
Fast forward to the prospect of a 50MP non-AA vs. a 50MP AA. Personally, I don't need 50MP ... but I do have a very strong preference toward a non-AA filter. My reasoning for this is that sharp / blur is an issue of contrast. By NOT having an AA involved, we are not reducing contrast @ capture, only to increase contrast in processing. The less contrast enhancement we impart, the less artifacts we risk inducing.
So ... non-AA incurs risk of moire' ... AA incurs risk of artifacts @ pushing contrast. As the pixels sizes become increasingly smaller, the issues of each become seemingly less detrimental. In order to incur moire' @ 50MP ... good grief, you are going to have to have some incredibly sharp glass and stellar technique to outresolve the sensor. OTOH, good grief, the diff @ 50MP AA vs. 50MP non-AA is going to be imperceptible to many as the contrast difference is mitigated by good PP and print sizes @ utility.
That being said ... as an ardent fan of non-AA ... unless you REALLY KNOW why you want an non-AA filter @ 50MP, then you probably don't really want a non-AA filter @ 50MP. At 14MP it can make quite a bit of diff. For most folks @ 50MP, they won't realize the diff. That said, I'd wait to see the comps from the cameras when they come out ... but in real world @ 50MP, FEW people will really warrant the diff @ non-AA vs. AA.
Of interest to note that some medium format digital cameras use removable AA filters, so they can utilize AA filter protection from moire' when shooting fashion, etc. and can remove the AA filter for non-pattern repeating subjects, such as landscape. Considering the limits of print, even an AA filter camera will likely outresolve paper on most days. So, on one hand you have your subject @ pattern vs. non-pattern. On the other hand you have lens resolving capability and print resolving capability ... all pieces of the puzzle regarding your decision for AA or non-AA. If you aren't that retentive about your shooting and processes involved ... skip the whole thing and just go with the AA filter like most folks.
Personally, I doubt my M645 glass will be sharp enough to outresolve a non-AA filter. Not sure about the now 100-400 5.6 II that is on my radar though. However, I won't be using a 100-400/5.6 II on fabric, so I expect to be safe. Conversely, if I do portraits with my M645 glass, I should be safe there also. IN one scenario, it is my subject matter that is preventing the moire' @ non-pattern repeating subject. In the other, it is the glass of choice inability to outresolve the sensor. Either way, moire' averted.
And, for those who shoot UWA, the resolving capability of UWA glass is typically much less than super long glass as the angles of incidence @ collimated light are significantly different to generate higher resolving capability in the longer glass. As a WA / UWA shooter ... outresolving a 50MP sensor is probably going to be a trick to achieve, even if you tried. For a birder, shooting long glass at a repeating feather pattern ... hmmmm @ non-AA for false color (but 50MP is really tiny).
For the OP ... if this has your head spinning ... borrow or rent a 50MP non-AA version and mount your sharpest lens. Shoot some fabric, some screens, some air conditioner grill vents, birds feathers, etc @ repeating patterns to try and induce some moire' / false color. Compare that to your shooting style and you can decide how the risk/reward tips your decision.
For me ... that's for me ... I'll suspect I'll always be a fan of non-AA (until I become a fashion photographer, anyway).
HTH
Edited on Feb 04, 2015 at 09:50 PM · View previous versions
Feb 04, 2015 at 08:59 PM
p.2 #8No low pass filter
Tough to say which way is 'right' since there are sound technical arguments as well as subjective preferences. So much about photography is about subjective matters.
Having shot with 18MP non-AA since 2010 and 24MP non-AA for about a year, there are definitely times when aliasing is an obvious problem, but less so with the 24MP sensor, which leads me to think it will be even less of a factor at 50MP in most 'real world' situations.
If you shoot a lot of people events, particularly formal events with fine patterned textiles, moire will appear. It will also surface in scenes containing certain kinds of manmade objects, such as brick walls, anything with a mesh (window sreens) or other regular repeating pattern. Having once shot some interiors on the 18MP sensor where most of the windows were screened, I had to stop down to f/16 to try to minimize the problem, intentionally misfocus and try different shooting distances. And moire was still a problem. At least with color moire, most PP software seems to handle it OK. It's when you end up with luminance moire that you're SOL, as illustrated in the earlier example.
My first digital was the 4MP 1D and I wonder how strong its AA filter was. In any case, at such low resolution, any kind of people event I shot ended up with cases of really nasty moire. As resolution increased, it hasn't been so bad, but I still see it at times with the 1DX, though it's usually mild compared to the other camera with the 18MP non-AA sensor. At 50MP though... I suspect the effect of the AA won't be that strong, so it will be a question of whether I want it for wider aperture people/events photography or can mitigate it in architecture type shots by stopping down sufficiently to induce diffraction. I have to admit, I do like a slightly crunch image, even if I know some of that detail is false, so long as it doesn't obviously look bad (definitely a subjective matter)... This was a matter of argument for some time over in the Alt forum's Leica M image thread, and perhaps also with the Ricoh GXR A12 module... and the general preference seemed to be in favor of crunchy, despite the technical arguments for AA filters.
But at least with Leica M, there are technical reasons AA was excluded, due to lens image quality degradation factors unique to that mount and the short exit pupil distance of many of the lenses. A situation that isn't relevant to most SLR lenses...
Having shot with 18MP non-AA since 2010 and 24MP non-AA for about a year, there are definitely times when aliasing is an obvious problem, but less so with the 24MP sensor, which leads me to think it will be even less of a factor at 50MP in most 'real world' situations.
If you shoot a lot of people events, particularly formal events with fine patterned textiles, moire will appear. It will also surface in scenes containing certain kinds of manmade objects, such as brick walls, anything with a mesh (window sreens) or other regular repeating pattern. Having once shot some interiors on the 18MP sensor where most of the windows were screened, I had to stop down to f/16 to try to minimize the problem, intentionally misfocus and try different shooting distances. And moire was still a problem. At least with color moire, most PP software seems to handle it OK. It's when you end up with luminance moire that you're SOL, as illustrated in the earlier example.
My first digital was the 4MP 1D and I wonder how strong its AA filter was. In any case, at such low resolution, any kind of people event I shot ended up with cases of really nasty moire. As resolution increased, it hasn't been so bad, but I still see it at times with the 1DX, though it's usually mild compared to the other camera with the 18MP non-AA sensor. At 50MP though... I suspect the effect of the AA won't be that strong, so it will be a question of whether I want it for wider aperture people/events photography or can mitigate it in architecture type shots by stopping down sufficiently to induce diffraction. I have to admit, I do like a slightly crunch image, even if I know some of that detail is false, so long as it doesn't obviously look bad (definitely a subjective matter)... This was a matter of argument for some time over in the Alt forum's Leica M image thread, and perhaps also with the Ricoh GXR A12 module... and the general preference seemed to be in favor of crunchy, despite the technical arguments for AA filters.
But at least with Leica M, there are technical reasons AA was excluded, due to lens image quality degradation factors unique to that mount and the short exit pupil distance of many of the lenses. A situation that isn't relevant to most SLR lenses...Feb 04, 2015 at 09:30 PM
p.2 #9No low pass filter
With wildlife (fur, feathers) I routinely get the 'false color' moire even with the D800. Not looking forward to a body without an AA filter when my D800 dies.
Feb 04, 2015 at 09:46 PM
p.2 #10No low pass filter
Christian H wrote:
With wildlife (fur, feathers) I routinely get the 'false color' moire even with the D800. Not looking forward to a body without an AA filter when my D800 dies.
Yeah. I see it even with the D7100 which has a smaller pixel pitch. Sometimes there are colored sparkles around the individual hairs.
EBH
With wildlife (fur, feathers) I routinely get the 'false color' moire even with the D800. Not looking forward to a body without an AA filter when my D800 dies.
Yeah. I see it even with the D7100 which has a smaller pixel pitch. Sometimes there are colored sparkles around the individual hairs.
EBH
Feb 04, 2015 at 09:49 PM
p.2 #11No low pass filter
rscheffler wrote:
at least with Leica M, there are technical reasons AA was excluded, due to lens image quality degradation factors unique to that mount and the short exit pupil distance of many of the lenses.
+1 @ somewhat akin to the angle of incidence UWA from other mounts.
at least with Leica M, there are technical reasons AA was excluded, due to lens image quality degradation factors unique to that mount and the short exit pupil distance of many of the lenses.
+1 @ somewhat akin to the angle of incidence UWA from other mounts.
Feb 04, 2015 at 10:21 PM
p.2 #12No low pass filter
Unsharp masking is an analog procedure involving film overlays: today, we can do it digitally. Another sharpening technology from the film era is called acutance, a chemical-based form of edge enhancement. Nothing new under the sun, the human eye likes edges.
As for anti-aliasing filters, I require my cameras to have them—although the D30 and 1Ds would alias like crazy under the right conditions. I shoot fabrics fairly often, and moire is an ever-present hazard. In any case, I do not appreciate false data...crunchy is for peanut butter, not photography.
Another sharpening technology from the film era is called acutance, a chemical-based form of edge enhancement. Nothing new under the sun, the human eye likes edges.As for anti-aliasing filters, I require my cameras to have them—although the D30 and 1Ds would alias like crazy under the right conditions. I shoot fabrics fairly often, and moire is an ever-present hazard. In any case, I do not appreciate false data...crunchy is for peanut butter, not photography.
Feb 05, 2015 at 04:06 AM
p.2 #13No low pass filter
rico wrote:
.... In any case, I do not appreciate false data...crunchy is for peanut butter, not photography.
.... In any case, I do not appreciate false data...crunchy is for peanut butter, not photography.
Feb 05, 2015 at 06:04 AM
p.2 #14No low pass filter
ben egbert wrote:
So you never sharpened the scans? It's a normal part of the digital workflow to compensate for the scanning process.
Some films like Velvia did their own chemical sharpening, and of course you've probably seen some photo prints made with unsharp masks.
EBHI never scanned or printed. Just project the slide when you get it back. Not that I am against pp, I just hate the fact that RAW's are so crippled out of the camera. I spend 30 seconds max per image sharpening so its only a psychological issue.
How are RAW files crippled out of the camera?
EB-1 wrote
ben egbert wrote
Hmm, maybe I need to go back to film if I want to avoid sharpening in post.Some films like Velvia did their own chemical sharpening, and of course you've probably seen some photo prints made with unsharp masks.
EBH
How are RAW files crippled out of the camera?
Feb 05, 2015 at 06:28 AM
p.2 #15No low pass filter
ben egbert wrote:
So you never sharpened the scans? It's a normal part of the digital workflow to compensate for the scanning process.
Some films like Velvia did their own chemical sharpening, and of course you've probably seen some photo prints made with unsharp masks.
EBHI never scanned or printed. Just project the slide when you get it back. Not that I am against pp, I just hate the fact that RAW's are so crippled out of the camera. I spend 30 seconds max per image sharpening so its only a psychological issue.
Digital files are inherently sharper than film with it's various colour layers (for slide film) and several lenses to go through for B&W and colour neg. Not to mention batch variations, drying marks & dust, slight discrepancies in neg carriers, slides popping in the mounts etc etc.
As someone who values sharpness as you do (i.e. above all else) you should be delighted by the new Canons
RAW files are a gift from heaven for me
EB-1 wrote
ben egbert wrote
Hmm, maybe I need to go back to film if I want to avoid sharpening in post.Some films like Velvia did their own chemical sharpening, and of course you've probably seen some photo prints made with unsharp masks.
EBH
Digital files are inherently sharper than film with it's various colour layers (for slide film) and several lenses to go through for B&W and colour neg. Not to mention batch variations, drying marks & dust, slight discrepancies in neg carriers, slides popping in the mounts etc etc.
As someone who values sharpness as you do (i.e. above all else) you should be delighted by the new Canons
RAW files are a gift from heaven for me
Feb 05, 2015 at 07:05 AM
p.2 #16No low pass filter
rico wrote:
Unsharp masking is an analog procedure involving film overlays: today, we can do it digitally. Another sharpening technology from the film era is called acutance, a chemical-based form of edge enhancement. Nothing new under the sun, the human eye likes edges.
As for anti-aliasing filters, I require my cameras to have them—although the D30 and 1Ds would alias like crazy under the right conditions. I shoot fabrics fairly often, and moire is an ever-present hazard. In any case, I do not appreciate false data...crunchy is for peanut butter, not photography.
+1 @ film > USM / acutance
With pixels the size of peanuts (1Ds), moire' would certainly be an ever-present hazard, particularly for fabric with good contrast lighting.
My dialogue with MaxMax included a comment that if you aren't getting moire' your lenses aren't sharp enough ... referring to the relationship between the lens & sensor. That's not to say that we should be striving to achieve moire' or false colors, but that in the issue of "the weakest link", whenever you have a lens that is capable of out resolving your sensor, moire' is always a potential. ... even with AA filters (albeit not as much) as others have also noted.
By the time we get to 50MP, the ability (or delta thereof) of many lenses to out resolve the sensor is quite different from their ability to outresolve a 1Ds sensor. (I dig those large pixels @ tonal transitions, btw.)
I'm curious though about the "crunchy" being referenced. Why is non-AA being associated to "crunchy"?. To me, that suggests over sharpening is being applied, or the wrong type of sharpening for the image structure. It makes me think that people are applying the sharpening that they use for an AA filter on a non-AA filter application. Sharpening is not a "one size fits all."
While a given lens / sensor / lighting combination can utilize a consistent amount of sharpening, changing any of these factors on the resolving contrast can warrant a change in your sharpening algorithm. How much sharpening you need for an indirect / diffuse lighting with a UWA lens on a D800 will be different from a direct sunlit super-tele on a D800E (which is not a true non-AA, btw) in order to bring each to a similar level of acutance ... i.e. diff lighting contrast, diff lens contrast, diff sensor contrast ... they cannot possibly produce the same results, sooc. So, applying the same USM to each cannot be optimal for both. Either one will be undersharpened, or one will be oversharpened from the utility of a same USM algorithm.
Part of the reason, I prefer the non-AA filter (or a very weak one) is that you do apply a different amount of sharpening (i.e. less) to achieve similar levels of acutance, etc. For those of us who do not appreciate "false data" ... well, all USM really is creating controlled false data in an effort to fairly represent the desired tonal transitions without creating undesired ones. The degree of over application renders the degree to which it "tells" on itself @ how much false data we are creating.
Applying USM to the point that it shows as crunchies (i.e. too much) is just a matter of showing your tells. Often times this can be due to applying global sharpening alone, when selective sharpening coupled with a lesser amount of global sharpening may be warranted to compensate for the varying structure within an image. While not everyone wishes to address their image sharpening selectively on a case by case basis of an image, that personal temperament could be an additional factor in evaluating whether AA or non-AA is a good fit for an individual workflow preference.
Starting from a non-AA filter baseline, the higher level of contrast (i.e. not being intentionally reduced @ capture) being recorded means that less "false data" application of USM is warranted to achieve the same level of "edges" that we are prone to like (for a given level), which in turn also helps keep the "crunchies" and other artifacts at bay ... as long as you didn't incur moire' / false color @ capture ... and don't try to apply AA filter level USM to a non-AA file.
Unsharp masking is an analog procedure involving film overlays: today, we can do it digitally.
Another sharpening technology from the film era is called acutance, a chemical-based form of edge enhancement. Nothing new under the sun, the human eye likes edges.As for anti-aliasing filters, I require my cameras to have them—although the D30 and 1Ds would alias like crazy under the right conditions. I shoot fabrics fairly often, and moire is an ever-present hazard. In any case, I do not appreciate false data...crunchy is for peanut butter, not photography.
+1 @ film > USM / acutance
With pixels the size of peanuts (1Ds), moire' would certainly be an ever-present hazard, particularly for fabric with good contrast lighting.
My dialogue with MaxMax included a comment that if you aren't getting moire' your lenses aren't sharp enough ... referring to the relationship between the lens & sensor. That's not to say that we should be striving to achieve moire' or false colors, but that in the issue of "the weakest link", whenever you have a lens that is capable of out resolving your sensor, moire' is always a potential. ... even with AA filters (albeit not as much) as others have also noted.
By the time we get to 50MP, the ability (or delta thereof) of many lenses to out resolve the sensor is quite different from their ability to outresolve a 1Ds sensor. (I dig those large pixels @ tonal transitions, btw.)
I'm curious though about the "crunchy" being referenced. Why is non-AA being associated to "crunchy"?. To me, that suggests over sharpening is being applied, or the wrong type of sharpening for the image structure. It makes me think that people are applying the sharpening that they use for an AA filter on a non-AA filter application. Sharpening is not a "one size fits all."
While a given lens / sensor / lighting combination can utilize a consistent amount of sharpening, changing any of these factors on the resolving contrast can warrant a change in your sharpening algorithm. How much sharpening you need for an indirect / diffuse lighting with a UWA lens on a D800 will be different from a direct sunlit super-tele on a D800E (which is not a true non-AA, btw) in order to bring each to a similar level of acutance ... i.e. diff lighting contrast, diff lens contrast, diff sensor contrast ... they cannot possibly produce the same results, sooc. So, applying the same USM to each cannot be optimal for both. Either one will be undersharpened, or one will be oversharpened from the utility of a same USM algorithm.
Part of the reason, I prefer the non-AA filter (or a very weak one) is that you do apply a different amount of sharpening (i.e. less) to achieve similar levels of acutance, etc. For those of us who do not appreciate "false data" ... well, all USM really is creating controlled false data in an effort to fairly represent the desired tonal transitions without creating undesired ones. The degree of over application renders the degree to which it "tells" on itself @ how much false data we are creating.
Applying USM to the point that it shows as crunchies (i.e. too much) is just a matter of showing your tells. Often times this can be due to applying global sharpening alone, when selective sharpening coupled with a lesser amount of global sharpening may be warranted to compensate for the varying structure within an image. While not everyone wishes to address their image sharpening selectively on a case by case basis of an image, that personal temperament could be an additional factor in evaluating whether AA or non-AA is a good fit for an individual workflow preference.
Starting from a non-AA filter baseline, the higher level of contrast (i.e. not being intentionally reduced @ capture) being recorded means that less "false data" application of USM is warranted to achieve the same level of "edges" that we are prone to like (for a given level), which in turn also helps keep the "crunchies" and other artifacts at bay ... as long as you didn't incur moire' / false color @ capture ... and don't try to apply AA filter level USM to a non-AA file.
Feb 05, 2015 at 07:41 AM
p.2 #17No low pass filter
Thank you everyone for your replies I've learned a lot. I've been reading canon rumors and it states that this 5DSR has the both capability so I asked in my understanding this properly does the 5DSR have a low pass filter and the cancellation of the low pass filter if that is so I would buy the 5DS R thank you in advance.Bill
Feb 05, 2015 at 10:53 AM
p.2 #18No low pass filter
shirozina wrote:
So you never sharpened the scans? It's a normal part of the digital workflow to compensate for the scanning process.
Some films like Velvia did their own chemical sharpening, and of course you've probably seen some photo prints made with unsharp masks.
EBHI never scanned or printed. Just project the slide when you get it back. Not that I am against pp, I just hate the fact that RAW's are so crippled out of the camera. I spend 30 seconds max per image sharpening so its only a psychological issue.How are RAW files crippled out of the camera?
They need to be sharpened. A raw file looks nothing like the actual scene. It is dull, lacks contrast, saturation and sharpness. At least with film, the starting point out of the camera was closer to reality. I could view my Kodachomes without doing anything to them. If I nailed focus with a good lens, it did not need sharpening and it certainly did not need any saturation. It did leave a lot to be desired in DR however.
ben egbert wrote
EB-1 wrote
ben egbert wrote
Hmm, maybe I need to go back to film if I want to avoid sharpening in post.Some films like Velvia did their own chemical sharpening, and of course you've probably seen some photo prints made with unsharp masks.
EBH
They need to be sharpened. A raw file looks nothing like the actual scene. It is dull, lacks contrast, saturation and sharpness. At least with film, the starting point out of the camera was closer to reality. I could view my Kodachomes without doing anything to them. If I nailed focus with a good lens, it did not need sharpening and it certainly did not need any saturation. It did leave a lot to be desired in DR however.
Feb 05, 2015 at 10:57 AM
p.2 #19No low pass filter
RustyBug wrote:
Another sharpening technology from the film era is called acutance, a chemical-based form of edge enhancement. Nothing new under the sun, the human eye likes edges.
As for anti-aliasing filters, I require my cameras to have them—although the D30 and 1Ds would alias like crazy under the right conditions. I shoot fabrics fairly often, and moire is an ever-present hazard. In any case, I do not appreciate false data...crunchy is for peanut butter, not photography.+1 @ film > USM / acutance
With pixels the size of peanuts (1Ds), moire' would certainly be an ever-present hazard, particularly for fabric with good contrast lighting.
My dialogue with MaxMax included a comment that if you aren't getting moire' your lenses aren't sharp enough ... referring to the relationship between the lens & sensor. That's not to say that we should be striving to achieve moire' or false colors, but that in the issue of "the weakest link", whenever you have a lens that is capable of out resolving your sensor, moire' is always a potential. ... even with AA...Show more →
Good reply. Of all the things we do in post processing sharpening is the first I would totally eliminate. With a sharp lens and enough resolution it ought not be required at all. The next would be color accuracy. My main problem with both of these is that I cant see sharpness or color well enough to fix it later. I need the camera to get it right.
rico wrote
Unsharp masking is an analog procedure involving film overlays: today, we can do it digitally. Another sharpening technology from the film era is called acutance, a chemical-based form of edge enhancement. Nothing new under the sun, the human eye likes edges.As for anti-aliasing filters, I require my cameras to have them—although the D30 and 1Ds would alias like crazy under the right conditions. I shoot fabrics fairly often, and moire is an ever-present hazard. In any case, I do not appreciate false data...crunchy is for peanut butter, not photography.
With pixels the size of peanuts (1Ds), moire' would certainly be an ever-present hazard, particularly for fabric with good contrast lighting.
My dialogue with MaxMax included a comment that if you aren't getting moire' your lenses aren't sharp enough ... referring to the relationship between the lens & sensor. That's not to say that we should be striving to achieve moire' or false colors, but that in the issue of "the weakest link", whenever you have a lens that is capable of out resolving your sensor, moire' is always a potential. ... even with AA...Show more →
Good reply. Of all the things we do in post processing sharpening is the first I would totally eliminate. With a sharp lens and enough resolution it ought not be required at all. The next would be color accuracy. My main problem with both of these is that I cant see sharpness or color well enough to fix it later. I need the camera to get it right.
Feb 05, 2015 at 11:01 AM
p.2 #20No low pass filter
.At least with film
Only if you shot slides, otherwise film looked nothing like reality when it came out of the camera.
Only if you shot slides, otherwise film looked nothing like reality when it came out of the camera.
Feb 05, 2015 at 11:06 AM