Capture One 6 NR much improved
I have been playing with the new upgrade, testing its NR capabilities on a number of very high ISO shots that I had previously worked on in C1-5. There is no question that the NR function is different and better. Gone are all of the white speckles that used to drive me crazy. There is still a bit more blotchiness than I would like, but much less than before. By experimenting with the luminance and surface noise sliders, I can get very acceptable results, whereas I could not do so with the previous version.
I have also done some quick head to head comparisons with the latest version of ACR, and I would say that C1-6 is a very strong challenger in the NR department. At 300%, ACR shows slightly smoother edges and fewer stray black pixels, but at 50-100%, it is hard to say that it is superior. I will be interested in hearing what other users think. Topaz DeNoise is still king of the hill, IMO, but for most purposes, I think that C1-6 will be able to produce very usable images on its own. Given the fact that the latest generation of sensors generate relatively little noise, there will be correspondingly less need to do NR in the post-conversion stage. For extremely noisy images, however, it will still be necessary. People are shooting at astronomical ISOs these days.
Great work, Phase One!
Rob
I have also done some quick head to head comparisons with the latest version of ACR, and I would say that C1-6 is a very strong challenger in the NR department. At 300%, ACR shows slightly smoother edges and fewer stray black pixels, but at 50-100%, it is hard to say that it is superior. I will be interested in hearing what other users think. Topaz DeNoise is still king of the hill, IMO, but for most purposes, I think that C1-6 will be able to produce very usable images on its own. Given the fact that the latest generation of sensors generate relatively little noise, there will be correspondingly less need to do NR in the post-conversion stage. For extremely noisy images, however, it will still be necessary. People are shooting at astronomical ISOs these days.
Great work, Phase One!
Rob
0
-
[quote="robgo2" wrote:
I do have to agree with Keith that some of the finest photographic artists that I know are amateurs. This is a subject that Brooks Jensen has written about in LensWork. Professionals generally can afford more expensive gear than all but a few amateurs, but what they do with that gear and how well they edit their own images is another matter altogether. I would not argue that a Phase One camera should be capable of higher IQ than any APS-C camera, all things being equal. Still, the IQ coming from the latest generation of APS-C cameras is pretty astounding and would have been unthinkable only a few years ago.
Rob
I agree that a professional label has little to do with the quality a person is likely to produce. So great shooters make great pictures period. A pro is no less likely to produce the highest quality than any other shooter.
There is no doubt that current digital technology is capable of astounding quality from all levels. A Canon G11 in capable hands can surely outshine a Kodak DCS660 from 2002 (original price $24000.00) but all this technology flows from the professional market where people pay huge sums for tech that a few years later is available to all. Good luck shooting with a non existent 7d in 2010 if there was no DCS660 or 1ds in 2002. So a little respect for all those crappy pro shooters who paved the way for the cameras you enjoy today.
Jeff H0 -
[quote="deejjjaaaa" wrote:
[quote="robgo2" wrote:
Here is one more ISO 3200 image. It's a challenging one--a fuzzy dark blue sweater with shiny yellow, gold, red and blue embroidery. All done in C1-6. Here are the NR settings: Luminance-20, Color-60, Single Pixel-50, Surface 50. At 100%, it looks reasonably clean, detailed, sharp and without major chock-a-block artifacts that I can see.
in fact the one before NR looks much better - the material of the sweater makes even ugly chroma noise look decent on it w/o any NR at all... 😂
I agree that the one before NR looked better0 -
[quote="Maggie12" wrote:
[quote="deejjjaaaa" wrote:
[quote="robgo2" wrote:
Here is one more ISO 3200 image. It's a challenging one--a fuzzy dark blue sweater with shiny yellow, gold, red and blue embroidery. All done in C1-6. Here are the NR settings: Luminance-20, Color-60, Single Pixel-50, Surface 50. At 100%, it looks reasonably clean, detailed, sharp and without major chock-a-block artifacts that I can see.
in fact the one before NR looks much better - the material of the sweater makes even ugly chroma noise look decent on it w/o any NR at all... 😂
I agree that the one before NR looked better
Is that because you like the color that the chroma noise imparts to the sweater? Of course, the point of the exercise was not to create a work of art, but rather to judge how effectively Capture One 6 can remove noise from a very noisy image without destroying detail and creating weird artifacts. I would appreciate it if people would base their judgements on those criteria.
Rob0 -
[quote="robgo2" wrote:
Is that because you like the color that the chroma noise imparts to the sweater? Of course, the point of the exercise was not to create a work of art, but rather to judge how effectively Capture One 6 can remove noise from a very noisy image without destroying detail and creating weird artifacts. I would appreciate it if people would base their judgements on those criteria.
Rob
It's not the color. It's this strange pixelated effect it created, I don't know how to describe it.0 -
[quote="Maggie12" wrote:
[quote="robgo2" wrote:
Is that because you like the color that the chroma noise imparts to the sweater? Of course, the point of the exercise was not to create a work of art, but rather to judge how effectively Capture One 6 can remove noise from a very noisy image without destroying detail and creating weird artifacts. I would appreciate it if people would base their judgements on those criteria.
Rob
It's not the color. It's this strange pixelated effect it created, I don't know how to describe it.
OK, so you like the way the noise looks. That's fine, but it does not address the question of C1-6's NR capabilities.
Rob0 -
[quote="robgo2" wrote:
[quote="Maggie12" wrote:
[quote="robgo2" wrote:
Is that because you like the color that the chroma noise imparts to the sweater? Of course, the point of the exercise was not to create a work of art, but rather to judge how effectively Capture One 6 can remove noise from a very noisy image without destroying detail and creating weird artifacts. I would appreciate it if people would base their judgements on those criteria.
Rob
It's not the color. It's this strange pixelated effect it created, I don't know how to describe it.
OK, so you like the way the noise looks. That's fine, but it does not address the question of C1-6's NR capabilities.
Rob
Correct, but only because the version without the noise looks strange ("pixelated"). When I look at the examples you posted later the one labeled ACR looks nicer. No noise and no strange "pixelation"0 -
[quote="Maggie12" wrote:
Correct, but only because the version without the noise looks strange ("pixelated"). When I look at the examples you posted later the one labeled ACR looks nicer. No noise and no strange "pixelation"
Huh. I don't see any pixelation. I would say that the ACR version is softer and flatter (less 3 dimensional.) This is especially apparent in the lettering at the bottom. In any case, what does seem quite clear to me is that C1-6 has a very respectable NR function.
Rob0 -
[quote="robgo2" wrote:
[quote="Maggie12" wrote:
Correct, but only because the version without the noise looks strange ("pixelated"). When I look at the examples you posted later the one labeled ACR looks nicer. No noise and no strange "pixelation"
Huh. I don't see any pixelation. I would say that the ACR version is softer and flatter (less 3 dimensional.) This is especially apparent in the lettering at the bottom. In any case, what does seem quite clear to me is that C1-6 has a very respectable NR function.
Rob
The letters look fine in either version. If I was just looking at the letters then I wouldn't even decide to remove any noise. It's fine the way it is. It's the dark part of the sweater that completely looses it's texture. It almost looks interpolated. If you asked me what it is I wouldn't be able to tell it's a sweater, all the other versions I can tell it's a sweater. I never said that I like noise. Only when compared to the version with NR I would rather have the noise in my image then loose the texture of the sweater. I also see strange lighter specks all over the dark part of the sweater0 -
[quote="Keith Reeder" wrote:
This whole argument comes down to do the artifacts you object to have a detrimental affect on a print?
Yes, they do. As I've explained several times now, the artifact either mucks up fine detail or requires such high Luma NR to deal with that the NR kills the detail anyway.
Keith,
This statement in your response gave me pause so I wanted to think about it and respond separately. Two of the oldest tricks in the book to increase apparent fine detail are to oversharpen or add noise. I find it odd that the artifacts you are railing against could muck up fine detail.
While aggressive noise reduction does in fact muck up fine detail. In the example Dejaaa posted this effect is clearly visible. The ACR version though sharpened is smoother but also softer and less detailed than the C1 version. You may not like the detail but it is there.
I have seen this particular demosaic artifact before. It was a common feature of the files from my old 1ds mk2 bodies. I had two of them and files from both contained this artifact when greatly magnified. I never did see this in any way affect a print though. In fact a 4'x6' lambda print still hangs in my studio from the 1ds mk2 displaying no evidence of the artifact. I really never was a C1 3 fan so all my files were run through Adobe Bridge into ACR.
Since we have this artifact appearing in two different Canon camera models separated by several years it is clearly a decision made by Canon in camera processing. The fact that ACR, Lightroom, Aperture, and C1 6 see and process it differently is not at all surprising. As Dejaaa's example demonstrates ACR leans toward smoother and C1 more detailed.
Finally, when I was making the transition to digital and the major catchphrase of the day was "accept responsibility for your own CMYK conversions" I attended a three day Advanced Color Seminar presented by Dan Margulis the worlds foremost color correction and prepress guru. We spent eight hours each day correcting horrendous files. When we got to pre press sharpening I was shocked at the levels applied in the name of preserving or enhancing fine detail. Levels far beyond any I had ever dreamed of applying before. That seminar changed the way I look at file prep and made me focus on how the file prints not how it looks on screen.
Jeff0 -
[quote="Jeff12" wrote:
[quote="Keith Reeder" wrote:
This whole argument comes down to do the artifacts you object to have a detrimental affect on a print?
Yes, they do. As I've explained several times now, the artifact either mucks up fine detail or requires such high Luma NR to deal with that the NR kills the detail anyway.
Since we have this artifact appearing in two different Canon camera models separated by several years it is clearly a decision made by Canon in camera processing. The fact that ACR, Lightroom, Aperture, and C1 6 see and process it differently is not at all surprising. As Dejaaa's example demonstrates ACR leans toward smoother and C1 more detailed.
Finally, when I was making the transition to digital and the major catchphrase of the day was "accept responsibility for your own CMYK conversions" I attended a three day Advanced Color Seminar presented by Dan Margulis the worlds foremost color correction and prepress guru. We spent eight hours each day correcting horrendous files. When we got to pre press sharpening I was shocked at the levels applied in the name of preserving or enhancing fine detail. Levels far beyond any I had ever dreamed of applying before. That seminar changed the way I look at file prep and made me focus on how the file prints not how it looks on screen.
Jeff,
I completely agree that C1's demosaicising pattern was a deliberate choice by Phase One's engineers to produce the particular image characteristics that are the signature of the program. It is what makes C1's images look so present and alive. I can barely perceive any artifacts even at 100%, except in extremely noisy photos. In fairness to ACR, I find that it produces tons of detail, but that the detail is rendered in a flat and busy looking fashion. Not bad, just blah.
What about Keith's claim that C1's artifacts show up in some of his bird prints, presumably done with inkjet printers? Is it possible that CMYK and inkjet printing produce different results? You must be using inkjet printers as well. He also claims that bird photography is a special category that requires the absolute cleanest, smoothest image files.
Rob0 -
[quote="Maggie12" wrote:
[quote="robgo2" wrote:
[quote="Maggie12" wrote:
Correct, but only because the version without the noise looks strange ("pixelated"). When I look at the examples you posted later the one labeled ACR looks nicer. No noise and no strange "pixelation"
Huh. I don't see any pixelation. I would say that the ACR version is softer and flatter (less 3 dimensional.) This is especially apparent in the lettering at the bottom. In any case, what does seem quite clear to me is that C1-6 has a very respectable NR function.
Rob
The letters look fine in either version. If I was just looking at the letters then I wouldn't even decide to remove any noise. It's fine the way it is. It's the dark part of the sweater that completely looses it's texture. It almost looks interpolated. If you asked me what it is I wouldn't be able to tell it's a sweater, all the other versions I can tell it's a sweater. I never said that I like noise. Only when compared to the version with NR I would rather have the noise in my image then loose the texture of the sweater. I also see strange lighter specks all over the dark part of the sweater
Maggie,
My guess is that you have zoomed in to the point that you are viewing the images at 200% or larger, which is irrelevant. On my computer, viewing at 100%, which is only slightly more relevant, there is no hint of interpolation in either of the images, and the C1 version has more detail and texture. (BTW, if there is any interpolation going on, it can only be due to jpeg compression. These images have not been re-sized by me.) Also, if you look closely, all of the white flecks show up in both versions, but they are sharper in C1's.
Rob0 -
Rob,
SInce I have seen the checkerboard pattern frequently in 1ds mk2 files I believe the pattern belongs to Canon and it is in the interpretation that C1 is producing an exaggeration.
I do use ACR and really like many of the features. I wasn't criticizing ACR for providing a smoother result it is just different.
All inkjet printers print CMYK and all ink spreads a bit when it hits the paper masking micro detail, smoothing noise, pixel borders etc... Fine detail presumably that which is above pixel resolution can be masked by using a fine art paper. So depending on the paper used and the magnification I suppose it is possible to enlarge the pattern enough that it is visible but then so are individual pixels. To my eye it is still a pattern that reads visually as noise so mucking up fine details is not something I would expect to or have seen.
Back when 6 megapixel was the DSLR standard it was common to use LAB sharpening and hammer the crap out of an image that contained grass for example to increase fine detail and reduce the digital feel. This was applied at the pre press stage and would make the picture look terribly noisy on screen. Once broken up into dots and printed though the grass on the page did have a more natural feel.
I really do not have an opinion on the quality standards of other genres. I do know that many large newspapers including the New York Times are shooting every day images as HD video so they can be printed in the paper and used a rich media source for web stills and video. I can say the most exacting IQ I have had to produce was for a local muralist who insisted on capturing canvas and brush stroke detail when his murals were copied for reproduction. I would shoot multiple 8x10 transparencies that would later be scanned and stitched together. Exposure, color, and contrast had to be matched and perfect for him to sign off. The bird shooters I know are very picky about IQ but do not have to produce the level of IQ required by my clients on a regular basis.
My experience with C1 and the Phase One products I have owned allows me to say confidently that the combo of P65+ and Rodenstock digital lF lenses on a 6x9 view camera is the highest digital IQ I have ever seen. Totally impractical for many of the people type projects I shoot so for me IQ is a sliding scale based on the project at hand.
By the way I like C1 6 and overall am very happy with the upgrade including NR.0 -
[quote="robgo2" wrote:
Maggie,
My guess is that you have zoomed in to the point that you are viewing the images at 200% or larger, which is irrelevant. On my computer, viewing at 100%, which is only slightly more relevant, there is no hint of interpolation in either of the images, and the C1 version has more detail and texture. (BTW, if there is any interpolation going on, it can only be due to jpeg compression. These images have not been re-sized by me.) Also, if you look closely, all of the white flecks show up in both versions, but they are sharper in C1's.
Rob
I am looking at what you have posted. Is that 200%?0 -
[quote="Maggie12" wrote:
[quote="robgo2" wrote:
Maggie,
My guess is that you have zoomed in to the point that you are viewing the images at 200% or larger, which is irrelevant. On my computer, viewing at 100%, which is only slightly more relevant, there is no hint of interpolation in either of the images, and the C1 version has more detail and texture. (BTW, if there is any interpolation going on, it can only be due to jpeg compression. These images have not been re-sized by me.) Also, if you look closely, all of the white flecks show up in both versions, but they are sharper in C1's.
Rob
I am looking at what you have posted. Is that 200%?
No, what I posted is 100%, but your browser can be zoomed in so that it will appear pixelated.
Rob0 -
[quote="robgo2" wrote:
[quote="Maggie12" wrote:
[quote="robgo2" wrote:
Maggie,
My guess is that you have zoomed in to the point that you are viewing the images at 200% or larger, which is irrelevant. On my computer, viewing at 100%, which is only slightly more relevant, there is no hint of interpolation in either of the images, and the C1 version has more detail and texture. (BTW, if there is any interpolation going on, it can only be due to jpeg compression. These images have not been re-sized by me.) Also, if you look closely, all of the white flecks show up in both versions, but they are sharper in C1's.
Rob
I am looking at what you have posted. Is that 200%?
No, what I posted is 100%, but your browser can be zoomed in so that it will appear pixelated.
Rob
My browser is not zoomed in and what you posted doesn't look good at all. Are you trying to convince me that there is something wrong on my end???? Sorry, I just don't like what I see. I am not zoomed in, I am not looking at 200%, I see strange artifacts and that's all.0 -
Having followed this thread with interest, I am curious to know what the ISO threashold is when the concerns about C1's NR capabilties kick in - are we just talking about ISO's of say 1600 or over? I appreciate it may depend upon the DSLR used but would think that cropping 7D images at high ISOs would be a pretty stern test of any NR software. Even if I am photographing birds with my 1D4, I try to keep the ISO at 1000 or preferably less. 0 -
My browser is not zoomed in and what you posted doesn't look good at all. Are you trying to convince me that there is something wrong on my end???? Sorry, I just don't like what I see. I am not zoomed in, I am not looking at 200%, I see strange artifacts and that's all.
OK, then we simply disagree. I don't see strange artifacts, and I think that it looks pretty darned good.
Rob0 -
[quote="Jeremy82" wrote:
Having followed this thread with interest, I am curious to know what the ISO threashold is when the concerns about C1's NR capabilties kick in - are we just talking about ISO's of say 1600 or over? I appreciate it may depend upon the DSLR used but would think that cropping 7D images at high ISOs would be a pretty stern test of any NR software. Even if I am photographing birds with my 1D4, I try to keep the ISO at 1000 or preferably less.
Clearly, I think that it will depend upon the particular camera used, the lighting, the exposure and the degree of cropping. In my testing so far, I think that C1-6 does a very decent job removing noise from images with light to moderate noise, meaning that it should be quite adequate for most images up to ISO 1600 from a 7D. For super-heavy, clumpy noise, I prefer Topaz DeNoise 5. It's a pretty amazing program, once you learn how to get the most from it. Also, remember that frequently, noise that you can see on the monitor does not show up in prints. So, just keep shooting at the lowest ISO that you can get away in any given situation, and try various NR solutions according to needs.
Rob0 -
I would like to share my experience and notices about CO6 NR characteristics as compared to LR3.
This examle was shot with Canon 5d mark II, iso 3200, exp.comp +1.5, color NR and tiny amount of Luma NR.
http://fotkidepo.ru/photo/361442/11297OSkmTcOfBT/blYSsowqgM/572910.jpg
As you can see LR3 version more unifom, more 'dotted' and more smoother shapes, has no color noise. CO6 version more detailed in contrast, but has little amount of color noise grains (if try to remove them, colors begins transit between objects - ex., blues from skirt goes to green blouse).
Look what if we add some Luma NR:
http://fotkidepo.ru/photo/361442/11297OSkmTcOfBT/blYSsowqgM/572911.jpg
LR3 version seems more blured, smoothed shapes, keeps fine grains but lose any details ... CO6 removes some fine noise, but keeps contrast details (which may look as large grains). In this case, the whole image resized to fit to screen (and maybe on print) looks more detailed.
So...
Color NR in CO6 worse then in LR3 (and many others Raw-convertors, as well), because it leaves some color noise grains or transit colors between colored objects, turns them grey.
Concept of Luma NR in CO6 better then in LR3, because it try to keep details while removes fine noise.
NR improvements in CO6 barely visible (as compared to CO5), and I realy loved if Phase One continue powerful NR improvements in next releases!0 -
NR improvements in CO6 barely visible (as compared to CO5), and I realy loved if Phase One continue powerful NR improvements in next releases!
I disagree that it is not improved over v.5, but agree that it needs more work. Hope that we will see real improvements in updates of v.6.
Rob0 -
I get similar results as SergNikitoni has posted and to me the ACR conversion looks much better when considering noise reduction (I don't care for the mottled look which takes away from the detail advantage that C1 has over Adobe -- and I can see that in my prints, that's why I object). Also from Rob's examples I'd pick the ACR but that's maybe beside the point. The point is that, coming back to the sentiment of Rob's original post and his last opinion, I do not see *any* perceptible improvements in NR when coming from v.5 to v.6. When I output an ISO 1000 file with no noise reduction and sharpening they look identical in both cases, and also when I use the exact same default values for NR and sharpening in v.5 and v.6 I get the same results. So I'm really surprised at the claims of "much improved" NR in ver.6. I don't see it on my files and I find it hard to believe that Rob's K-7 files would behave so much differently from my K10D files.
So here's a challenge for you, Rob, and I mean it in a non-aggressive way -- could you post 100% crops of your conversions of the same file with the same NR and sharpening and clarity settings from both v.5 and v.6 of C1?0 -
[quote="sankos" wrote:
I get similar results as SergNikitoni has posted and to me the ACR conversion looks much better when considering noise reduction (I don't care for the mottled look which takes away from the detail advantage that C1 has over Adobe -- and I can see that in my prints, that's why I object). Also from Rob's examples I'd pick the ACR but that's maybe beside the point. The point is that, coming back to the sentiment of Rob's original post and his last opinion, I do not see *any* perceptible improvements in NR when coming from v.5 to v.6. When I output an ISO 1000 file with no noise reduction and sharpening they look identical in both cases, and also when I use the exact same default values for NR and sharpening in v.5 and v.6 I get the same results. So I'm really surprised at the claims of "much improved" NR in ver.6. I don't see it on my files and I find it hard to believe that Rob's K-7 files would behave so much differently from my K10D files.
So here's a challenge for you, Rob, and I mean it in a non-aggressive way -- could you post 100% crops of your conversions of the same file with the same NR and sharpening and clarity settings from both v.5 and v.6 of C1?
Sankos,
No can do. C1-5 has been removed from my computer. Why don't you provide the examples yourself?
BTW, I have never claimed that demosaicising has changed from v.5 to v.6, although Phase One does say that IQ has been improved, whatever that means. What I did say is that when I took noisy images previously edited in v.5, I got better NR in v.6. The settings were not identical, but in both cases, they were the best that I could achieve. I cannot now go back to re-do them in v.5.
I still contend (and have no doubt) that for images with light to moderate noise, C1 does a very acceptable job of NR. It is not as smooth as ACR, but it is sharper and more detailed. Some images are harder to treat than others, and on some of those, C1 can produce blotchiness. For those images, I would use Topaz DeNoise, not ACR. As SergNikitoni demonstrated, ACR really does kill detail in the process of producing its nice, fine grain. One should not lose sight of that fact.
Rob0 -
[quote="SergNikitoni" wrote:
I would like to share my experience and notices about CO6 NR characteristics as compared to LR3.
how about posting a link to the raw file + development parameters ?[quote="SergNikitoni" wrote:
color NR and tiny amount of Luma NR.
how much ?[quote="SergNikitoni" wrote:
Look what if we add some Luma NR:
how much ?
---0 -
[quote="robgo2" wrote:
As SergNikitoni demonstrated
he demonstrated nothing... just some resized .jpgs, from some unknown raw file, developed w/ some unknown parameters.0 -
[quote="SergNikitoni" wrote:
I would like to share my experience and notices about CO6 NR characteristics as compared to LR3.
This examle was shot with Canon 5d mark II, iso 3200, exp.comp +1.5, color NR and tiny amount of Luma NR.
Does anyone else find it odd that we are not only judging but criticizing C1 6 noise reduction by using only extreme examples. I understand the temptation but if I make a decision to shoot at 3200 iso and push the exposure 1.5 stops it is because I am forced to just to get a picture. Noise is a part of that equation regardless what NR software I use. I would choose Topaz denoise, print the photo on a rag paper to further mask noise and consider myself ahead of the game to get a cool photo out of that type situation.
Rob has been saying all along that he finds the NR more effective with more moderate images.0 -
how about posting a link to the raw file + development parameters ?
Post your own examples and raw files. I presume that you own a camera.
Rob0 -
[quote="Jeff12" wrote:
[quote="SergNikitoni" wrote:
I would like to share my experience and notices about CO6 NR characteristics as compared to LR3.
This examle was shot with Canon 5d mark II, iso 3200, exp.comp +1.5, color NR and tiny amount of Luma NR.
Does anyone else find it odd that we are not only judging but criticizing C1 6 noise reduction by using only extreme examples. I understand the temptation but if I make a decision to shoot at 3200 iso and push the exposure 1.5 stops it is because I am forced to just to get a picture. Noise is a part of that equation regardless what NR software I use. I would choose Topaz denoise, print the photo on a rag paper to further mask noise and consider myself ahead of the game to get a cool photo out of that type situation.
Rob has been saying all along that he finds the NR more effective with more moderate images.
Jeff,
You are absolutely right, but testing gear and software at the extremes is how the game is played on the internet. Just to prove my point, I am going to post a real stress test, both for my camera and Capture One. Below are 100% crops from a close-up of some blue jeans. They were shot underexposed with a Pentax K-7 at ISO 3200 and boosted one full stop in raw, making them the equivalent of ISO 6400, which is way beyond what I would ever shoot with this camera.
C1-6, No NR-100%
http://gallery.photo.net/photo/12089555-lg.jpg
ACR-100%
http://gallery.photo.net/photo/12089553-lg.jpg
C1-6-100%
http://gallery.photo.net/photo/12089554-lg.jpg
Now, I think that almost everyone will agree that the ACR version is cleaner, even if a bit flatter and softer. Nevertheless, the C1 version is respectable, and when viewed at 50% enlargement, which is still big but closer to a real world viewing size, it looks slightly grainy, but overall quite decent. I think that most people would find it perfectly acceptable.
C1-6-50%
http://gallery.photo.net/photo/12089592-md.jpg
And I have no doubt that if it were to be printed at a reasonable size, it would actually look quite good.
Of course, this is just an exercise. In my real photographic life, I would perform NR with Topaz DeNoise in an extreme case such is this. I'm sure there will be those who will jump all over this demonstration as proving nothing, but that is fine with me. I'm having fun and becoming more convinced that C1-6's NR is really quite usable.
Rob0 -
[quote="deejjjaaaa" wrote:
[quote="SergNikitoni" wrote:
I would like to share my experience and notices about CO6 NR characteristics as compared to LR3.
how about posting a link to the raw file + development parameters ?[quote="SergNikitoni" wrote:
color NR and tiny amount of Luma NR.
how much ?[quote="SergNikitoni" wrote:
Look what if we add some Luma NR:
how much ?
---
I do not think that exact values of NR parameters from different software is important. In this comparition I tried to use such values to get more similar looking from both converters and to look how NR algorythms works.
And this is 100% crops - not resize!
My aim is to analyze CO's noise recuction characteristics to improve it, because I think that this is the weakest point of this REALY greate software!
Next, let's find the difference between CO5.2.1 and CO6.0 NR...
In this case Luma NR=5, Color NR = 44 (42 for CO6), Single Pixel = 51, no adv. NR, 100% crops
http://fotkidepo.ru/photo/361442/11297OSkmTcOfBT/blYSsowqgM/573084.jpg
As we can see, Color NR in CO6 becomes little more agressive with the same NR values.
Some color noise grains '3' is still visible in ver 6 result, but color fade and transition present yet ('1', '2').
So, color NR worked better in ver 5 ☹️ - less color noise and more color details.
Luma NR seems works the same.
Sad but I can't see any NR improvements. Maybe this improvements works fine with lower ISO shots.
Robgo2,
Why sould I use other NR software (Topaz, NoiseWare, NoiseNinja...) to eliminate noise? RAW-files has more information then jpegs and tifs, that's why CO has potential ability to eliminate pattern noise better than post production SW.
P.S.
I have found some improvements! In the same image:
http://fotkidepo.ru/photo/361442/11297OSkmTcOfBT/blYSsowqgM/573096.jpg
This big green-magenta spots on the wall in ver 5, cannot be removed in ver 5 with any Color NR value.
So this some kind of improvement, which can be seen even in resize for fit on screen. 😊0 -
[quote="robgo2" wrote:
You are absolutely right, but testing gear and software at the extremes is how the game is played on the internet. Just to prove my point, I am going to post a real stress test, both for my camera and Capture One. Below are 100% crops from a close-up of some blue jeans. They were shot underexposed with a Pentax K-7 at ISO 3200 and boosted one full stop in raw, making them the equivalent of ISO 6400, which is way beyond what I would ever shoot with this camera.
And I have no doubt that if it were to be printed at a reasonable size, it would actually look quite good.
Of course, this is just an exercise. In my real photographic life, I would perform NR with Topaz DeNoise in an extreme case such is this. I'm sure there will be those who will jump all over this demonstration as proving nothing, but that is fine with me. I'm having fun and becoming more convinced that C1-6's NR is really quite usable.
Rob
Rob,
Test away, understanding the strengths and limitations of any tool is important. Any exercise that allows the photographer to gain knowledge is surely useful and worthwhile.
In this thread though participants were using this one narrow but important parameter to condemn the whole package. Even though the testing was extreme and did not take into account performance across the spectrum of supported cameras and situations. For me that is wrong and to be challenged.
The best NR tool I know of is our own brain and the thoughtful application of knowledge and experience to shape the picture we want see. It is the rejection of this idea in favor of laying blame on C1 that drew me to comment on this thread.
Jeff0 -
Why sould I use other NR software (Topaz, NoiseWare, NoiseNinja...) to eliminate noise? RAW-files has more information then jpegs and tifs, that's why CO has potential ability to eliminate pattern noise better than post production SW
.
The key phrase here is: "CO has the potential ability..." The fact of the matter is that Capture One has not yet reached its full NR potential, and there are other programs that do a better job. I certainly prefer to do NR in the raw stage, and I am very confident that C1-6 will suffice for most images with light to moderate noise. But the Phase One engineers need to keep working to improve the NR function.
Edit: I am at work now, and on my crappy monitor, I can barely see the samples that you posted. I will try to look at them again when I get home tonight.
Rob0
Post is closed for comments.
Comments
127 comments