Raw files converted to tiff look different when viewed in Capture One
Raw files converted to tiff (when using the Edit With function) look different when viewed in Capture One.
Here's an example – I realize it may be a little difficult to see, but the whites in the tiff are less bright and some parts are more yellowish (tiff to the left):
Converted with these settings:
Is this how it's supposed to be?
-
Does it make any difference if you choose sRGB instead of Adobe RGB?
Ian
1 -
Thanks for the suggestion.
It makes no difference as far as I can see.
Thomas
0 -
OK. It was worth checking.
Ian
0 -
The greens of the trees. Details (sharpness or microcontrast) e.g. trees, the clock.
You compare the tiff in C1 vs. the raw in C1? Softproof? I think you should. Which recipe for soft-proofing? Not sure the preview of the tiff file in C1 should be the benchmark, better to compare the raw preview in C1 with soft-proofing set to the export recipe vs. the exported tiff in a color managed app like PS. But still, I think, minute differences can be expected.
0 -
Exactly, the greens of the trees, details and sharpness.
I compared the two in Capture One with soft proofing off. When switching back and forth between them the difference was very apparent. I had just made some adjustments in Photoshop (though not to the tiff file itself) and wanted to compare the result to the original, and realized that both the tiff and the Photoshop version of it looked different from the raw file.
I assumed there would be some difference, but expected it to be much smaller than this.
I'll see what it looks like with soft proofing tomorrow.
0 -
Yes, John, and even other (in gamu) colors would be converted into the target color space, depending on the rendering intent. But the rendering intent determines the conversion during the actual export as well as the preview, if softproofing with that color space is enabled, so theoretically this should be shown as equal colors in C1.
There are however tools in C1 which are not applied during the raw preview during some zoom levels, e.g. NR (and maybe sharpening?), which only kicks in at about 66% or so. I think (not sure right now) that NR is enabled during softproofing even at fit to screen zoom level. For critical comparison, 100% zoom could be used, but of course that is not always how you want to look at images during editing...
Let's see what Thomas findings tomorrow will reveal to him.
1 -
Soft proofing makes no immediately apparent difference. I've tried different settings.
I've found that when I export the two images, either as tiff or as jpg, there's not much difference between them when viewed outside of Capture One; there may be a small difference, but it's nowhere near what I see in Capture One.
0 -
First, view the image at 1:1 magnification to compare colors. There is a color shift depending on the noise and zoom factor in C1 (This is the same RAW file at both 100% and 33% zoom).
Second, use the ProPhoto profile and 16-bit depth when exporting a TIFF to Photoshop. Ensure that both Photoshop and C1 retain the color space information (CMM settings, Adobe). Also, make sure that the edited TIFF file has the correct input profile.
Consider keeping a consistent color space throughout your workflow. You might want to try using ProPhoto or Adobe RGB.
It's important to note that the debayering, sharpening, and denoise algorithms in Capture One appear to be influenced by the zoom scale for reasons that are not entirely clear (may be speed). Remember to examine colors closely in 1:1 view to detect any shifts. Color shifts can occur depending on the noise present in the RAW file in C1, and this can vary with the zoom factor.
I hope I could help. Best regards, Andi.0 -
Thanks for the suggestion, Andreas, I will try that.
0 -
I've heard about Prophoto / ROMM RGB before.
"ICC makes no representations about the suitability of this profile for any purpose."(ROMM)
(https://www.color.org/chardata/rgb/rommrgb.xalter)I use AdobeRGB(1998) when exchanging TIFF files with Affinity Photo. The reason is that I have not yet found sufficient evidence / official advice which color space is best suited for exchanging TIFF files between applications. But maybe there is...?
Maybe I need to read again through Bruce Lindbloom's site:http://www.brucelindbloom.com/index.html?WorkingSpaceInfo.html
It should be noted that besides the gamut of a color space (the color range) also gamma, white point, slope limited or not, to name a few, and small but possbile quality losses during conversion and different implementations (handling, assumptions of the software) of different software applications can play into the already complex matter.
If I would exclusively print my images (no web, no long-term storage considerations with the "widest profile available") then I assume eciRGB would be my choice (unless the image has a lot of blue/magenta).
Anyway, 16 bit is a must for big color spaces.
Comparison of spaces:
https://support.captureone.com/hc/en-us/articles/360002671658-Comparison-of-spaces
ProPhoto RGB:
https://support.captureone.com/hc/en-us/articles/360002671798-ProPhoto-RGB
AdobeRGB:
https://support.captureone.com/hc/en-us/articles/360002658857
ECI RGB:
https://support.captureone.com/hc/en-us/articles/360002671838-ECI-RGB-v2
sRGB (includes informatinon about sRGB variants, which you probably didn't know):
https://support.captureone.com/hc/en-us/articles/360002671738-sRGB
EDIT:
There is tons of very detailed information (beyond the short info in the C1 links above) in the web, if someone knows a reliable and understandable source for information as to which space to use for what purpose ...please jump right in...
0 -
I suppose the acid test is whether you (or anyone else) can actually see a difference in the final print.
Ian
0 -
Yes, short-term.
If I want to keep a portfolio which includes e.g. tiff files and not just the raws plus raw converter settings I would like to use a colorspace which in the best case is suitable for future evolutions of printers and monitors, in addition to the present satisfaction.
EDIT: My reply reminds me to visit again the archival organisations websites to see what is best practise and recommended. C1 Cultural Heritage, looking at you too.. :-)
0 -
Apologies - this is quite straightforward. If you aim to edit a picture while preserving as much information as possible for the editing process, including the possibility of making minor adjustments later for printing or publishing in various media, it's essential to use a wide color space in 16-bit to prevent color distortion or clipping. In this context, proPhoto RGB with a gamma of 1.8 excels.
https://support.captureone.com/hc/en-us/articles/360002483178-Color-spacesIn Capture One, you have the option to select the ICC profile within the recipe. The chosen recipe determines how colors are processed, and under "Preset > Color Rendering," you can specify how colors are translated between different color spaces. https://helpx.adobe.com/photoshop-elements/kb/color-management-settings-best-print.html
The proofing function includes the output sharpening and resolution set in the recipe. However, if you believe this would prevent the color shifting issue explained in the previous post?. If you set the output resolution to match the display resolution. Unfortunately, the problem still persists; now, at 200%, the colors are shifting. Now I have wrong colors at 100%... right ones at 200% (crazy Sh**)
So, even if your plan is to archive your portfolio, it's still advisable to go with proPhoto RGB.
However, when it comes to reproduction, if your goal is not to recover highlights or details in dark areas but rather to work with data suitable for printing on paper or display on a monitor, then ECI RGB v2 or Adobe RGB would be the appropriate choice.
For the sake of completeness, I would like to mention that printing with ICC profiles in Capture One on Mac has not been functional for several years, and there is still no solution to this issue as of now.
"Thanks for contacting us. The issue you've described is a known problem that our Research and Development (R&D) teams are actively working to resolve. We are committed to providing a solution as soon as possible, although I cannot provide a definite timeframe for when it will be available. We will keep you informed of any developments."
So, perhaps the way Capture One processes color is somewhat overrated, as for printing, you have to use a different application anyway. I had major issues with some sharp Sigma lenses. I initially thought they had a problem with color rendering in soft areas of an image. However, a year ago, I came to the realization that it had nothing to do with the lenses themselves; rather, it was the way Capture One renders colors, which varies depending on the zoom factor for display. I think that the debayering, noise reduction, and sharpening processes are only fully executed when the zoom level is above 50%.
These are my findings on this topic. I still hold hope that one day printing will be possible again.
Andreas0 -
Andreas,
Thanks for link, I have seen and even commented it 4 years ago.:-)
So, this would be my summary:
If the purpose of your imagery is not only printing but long-term storage with unknow purpose on unknown hardware in the future, then ProPhoto RGB for all intermediate files, provided the applications can use it, either as a working space which would be ideal, or at least can convert it to their own internal working space.
Conversion from C1 internal working space to ProPhoto gamma 1.8 in the file, then converting in the target app to its internal working space (if it doesn't use Prophoto. PS does I think, Affinity, Gimp etc. I don't know), then conversion back to C1 internal working color space in case you make a round trip. All these conversions don't harm the colors, at least not those you can see on your state of the art monitor today (which cannot show all ProPhoto colors) or on your state of the art printer/paper.
If you print, only the last step would be AdobeRGB or ECI RGB (or jpg sRGB), depending on the capabilities of your printing service. (Or printing software, if you print yourself but not directly from C1).If you only print, no long-term storage for future purposes, then going directly to AdobeRGB or ECI RGB in the exported file could be better.
Web is sRGB 8bit currently, a little out of the discussion here I think.
In short: Prophoto for exported files to other apps, if possible, except for printing (AdobeRGB or ECI), sRGB for web.
I would use ROMM RGB as this is said to be equivalent to ProPhoto and can be downloaded from a trusted source (I don't have PS which ships ProPhoto).
Rendering Intent in C1 prefences should not matter too much for ProPhotoRGB, assuming C1 working space gamut is not wider than ProPhoto. But we don't know the C1 working space. So, perceptual rendering intent? It does matter for the other color profiles mentioned here and can be different for different printers/papers.
Zoom levels in C1:
I remember it from earlier years, and I repeated the same in this thread but it isn't true. I just checked C1 for Windows build 15.2 (100% and below) and sharpening and NR are applied to the preview at all zoom levels. Maybe this was changed in an earlier version, or it had to do with preview size or graphics drivers.
0 -
Retouching/Manipulation
Image files that are intended to receive intensive retouching and manipulation can benefit by being processed and output in 16-bit to ProPhoto RGB, which is a larger color space than Adobe RGB.(from that C1 link Andreas provided)
On the other hand, if I am doing only basic stuff with minor effort in external applications like Affinity and not intensive manipulation, and my target is either sRGB for web of AdobeRGB/sRGB for my printing service anyway, and for long-term usage e.g.with super duper printers I would start with the raw files incl. C1 settings again (if I am still around with C1), using the printer profile for soft-proofing (I like to restrict myself for tonal and color work to only one application) and repeating the basic stuff in Affinity, why should I risk putting ProPhoto with its very wide gamut of colors into the mix. I did not yet observe it yet, but theoretically the wider the gamut the higher the risk for banding, as you only have max 16bit integer available per color channel, regardless of profile. (Well, darktable has floating point). The wider the gamut / more colors but same number of numerical representations means bigger steps between the colors, one root cause for banding for e.g. AdobeRGB 8bit when manipulating colors.
Maybe 16bit integer is sufficient to prevent banding with ProPhoto?
P.S. btw, with DXO Photolab, all sharpening is gone below 75% zoom level.
0 -
Hello BeO!
I don't want to be impolite, but honestly, I don't even know where to begin, but I suspect you might feel the same way. And frankly, it doesn't matter much to me; I'm not on a missionary trip here, and it's unlikely we'll come to an agreement.
This image shows the same RAW image with identical settings at 100% and 25%, alright. I believe you can notice the difference in color. On Mac, there are more options available when it comes to dual display compared to Windows. I used to have a second viewer with a smaller version of the image I am working on to get an idea of how it would appear on an iPhone or as a thumbnail. And I wanted to draw attention to this issue, specifically that the colors in the small version do not match the ones in the larger one.
And here's an export of the same RAW file in 16-bit Adobe RGB at 100% / 50% / 25% sizes using Lightroom Classic, Capture One, and DXO PhotoLab. In Capture One, the colors vary depending on the export size. I've presented the RGB values in 8-bit for easier reference. Perhaps you could try it out for yourself.
"Working Color Space / Garmut" In Capture One, there isn't a working color space like in ACR (Photoshop / Lightroom). As far as I know, in ACR, the RAW data is always converted to proPhoto RGB. In Capture One, it transitions from the input color space to the output color space. I presume that the underlying mathematics are not restricted and involve 32-bit floating-point precision, although this is speculative. You can find more information about Capture One's color gamut here: https://support.captureone.com/hc/en-us/articles/360002861957-Capture-One-color-gamut
I believe that this is where most of the misunderstandings regarding proofing in Capture One originate. Affinity Photo operates in 32-bit linear mode. In this context as well, I would recommend using a larger color space to preserve information and prevent clipping. Nevertheless, I am not an expert in Affinity.You're consistently drawing comparisons between different things. One aspect pertains to color reproduction, while the other concerns how to retain information needed for editing. Regardless of your perspective, the fact remains that it is advantageous to preserve information, even if it is not required for reproduction, or when it cannot be accurately represented at all. This is particularly relevant when considering future applications, such as HDR-Displays (not HDR Tone Mapping).
You mentioned that the proPhoto RGB color space might be too extensive and potentially lead to banding even with 16-bit. From my viewpoint, 16-bit, which corresponds to 2^48, should handle colors without introducing banding and with minimal interpolation losses when transitioning between color spaces or gammas, L*, sRGB a.s.o.
From my point of view.
Best regards Andreas0 -
Hi Andreas,
No worries, I'm here to discuss, exchange and learn new things, perspectives and experiences. And coming to an agreement is not mandatory.
First, thanks for the examples, much appreciated, likely some work involved with it.
Question: Your first screenshot is with soft-proofing with which profile?
I hope I can do a similar test (as your export to files test) tomorrow, especially to see if I encounter color shifts in different export file resolutions with AdobeRGB too. I see them in your screenshots.
Question for my understanding: You actually exported to file (not softproof) and then opend them in PS and used the PS color picker to show the RGB values?Did you test different zoom levels of a raw file with regards to sharpening and NR kicking in and can confirm my recent observation that the tools are enabled all the time?
I fully agree that image information should be preserved as good as possible, I am just not sure though how to achieve that. I think you agree that color space conversion can and in many/some cases will destroy information? Whenever one converts a wider color space into a smaller color space, this seems quite obvious, as colors out of the source gamut will be squeezed into the smaller destination gamut, depending on the rendering intent in-gamut colors will be shifted ("perceptual") to make space for out-of-gamut colors, or out-of-gamut colors will be put on the destination border, hence eliminating gradient information relative to other pixels already at that border ("relative"). This is an argument for ProPhoto as it is wider as e.g. AdobeRGB.
However, there are also other conversion issues which eliminate color information of an image, and here I refer again to Bruce Lindbloom. According to him, actual (not theoretical) conversion from an RGB space to Lab (which contains all possible colors), will destroy colors because they are "squeezed" into the 16bit integer implementation of Lab. Sound counter-intuitive but this is due to the "coding efficiency" of Lab (in 16bit integer) of only 35%. (If I interpret him well).
http://www.brucelindbloom.com/index.html?WorkingSpaceInfo.html
Another interesting observation from the table relates to native Lab encoding. The established methods of integer encoding of Lab color (Lab TIFF, ICC, Photoshop) will clip some of the Lab Gamut. But even more devastating than that is the gross coding inefficiency (only 35%). This means that nearly two-thirds of Lab coding space is wasted on colors that do not even exist. This may be seen here. This inefficiency "squeezes" real colors tightly together, resulting in possible quantization losses. So converting an image into Lab for the purposes of applying a color correction in Photoshop can severly reduce the number of unique colors in your image. This is discussed further here. Whether this is a significant loss depends on the particular situation, but you should at least be aware of it.
ProPhotoRGB has almost all colors of Lab and a much better coding efficiency of 87%, but still information can get lost due to that "waste" of 13% which is responsible for other colors to be squeezed into the 87%. If I interpreted him well and my conclusions are correct.
Adobe has only 50% Lab gamut coverage but 100% coding efficiency, hence only losses due to perceptual rendering conversion (color shifting and squeezing).
If I plan to print then I need to live with 50% of Lab colors (AdobeRGB) intersected with the printer/paper gamut, and with the conversion to AdobeRGB anyway. So I am not sure that introducing a potentially "lossy" intermediate conversion to Prophoto will result in a better or worse or equal quality print, especially if I am not manipulating a lot in PS/Affinity.Don't read my wrong here, I don't want to be impolite either:
Is your workflow based on an assumption and only backed by the fact that ProPhoto has the wider gamut and the linked C1 article states it, or is it backed by other sites and/or own comparisons, that a conversion to a file using Prophoto preserves information better and also yields better web images or prints as the final output? How much manipulation do you do in PS?One of the more interesting discussions here in the forum, imo.
Best regards
BeO0 -
Hallo BeO!
My work, workflow and knowledge
Regarding my work and knowledge in Photoshop, it primarily involves exposure bracketing for interior architecture or product photography, often requiring extensive blending and detail restoration. I also pursue macro and wildlife photography as a hobby.My workflow often involves starting in Capture One for initial adjustments. Then, I move to Photoshop for denoising with Topaz and retouching or layer blending. After that, I go back to Capture One for final color grading and fine-tuning, also proofing for printing ... printing is done in a different app. I find proPhoto to be crucial in this workflow as it helps prevent individual color channels from clipping, ensuring more space to work.
However, what's perhaps more relevant is my extensive experience in color corrections and format conversions, particularly in the context of film and video production. I've also read a lot about it, but I must admit that in recent years, I've been more focused on HDR, ACeS, BT.2020, and similar topics.
Your quantization losses - theory
I can't follow your conclusion about quantization losses. They are understandable in an 8-bit workflow but not in 16-bit. I agree with you that if you're just removing some blemishes and imperfections in Photoshop and don't intend to make any more changes to exposure or colors, or if you don't need to convert to a different output color space, then using proPhoto RGB doesn't make sense. I find L* / eciRGB very useful, especially when it comes to printing and maintaining brightness consistency between the monitor and the print.
I tried a basic, primitive approach to test this. I assigned the Adobe RGB profile to the file from Bruce Lindbloom, with 16.777.216 different pixels. (as there are too many different profiles with sRGB). Then, I converted it to 16-bit proPhoto and imported and converted it as a layer back into 8-bit AdobeRGB. I identified the differing color values using the difference layer and then counted the exact same pixels. Only 0.09% of the pixels that differed had a Delta E ranging from 0,2833 to 0,8036. 99,0968% of the pixels had the same value.
On the left, it goes from 8-bit Adobe RGB to 16-bit proPhoto and then back to 8-bit Adobe RGB. 16,625,689 color values remained the same 16630396 are unique.On the right, it goes from 8-bit Adobe RGB to 8-bit proPhoto and then back to 8-bit Adobe RGB. Only 5,475,306 color values remained the same.
This primitive test doesn't contradict the statements made by Bruce Lindbloom or the ECI. The argument about quantization losses always pertains to 8-bit to 8-bit scenarios and overlooks the fact that in my workflow, I transform data from point A to point B, making the invisible visible and introducing changes (Coding Efficiency is not relevant).
Color-Shift in Capture One
Again... Proofing profiles for Capture One becomes irrelevant, since it's about going from source to target. Proofing also takes into account output sharpening and final resolution. The described problem has nothing to do with the output color space. You could simply reread the post and try to understand it, or perhaps it's my shortcomings in expressing myself.
1: The RAW image at 100% in another viewer window (I think the trick may not be possible on Windows). 2: The same file / variant, scaled at 25%. 3: The file re-imported as a TIFF into C1 - scaled without a color shift.
The color-shift effect only occurs with RAW files that are very noise or pushed very far. Hence my suggestion to Thomas Kyhn regarding the 1:1 view. This could also be a cause of the color shift.
You can find the Files here, have fun!
https://jenud.synology.me:5001/sharing/yFbTPFFQf
Pixel-Counter: https://townsean.github.io/canvas-pixel-color-counter/
Test Images: http://www.colormanagement.org/en/testimages.html
Best
Andreas0 -
Hi Andreas,
Thanks.
ProPhotoRGB
I've read the posts several times, and the article from Bruce too.
The argument [from Bruce] about quantization losses always pertains to 8-bit to 8-bit scenarios.
He wrote about integer implementation and, if I interpret him right, blamed it as the root cause for the quanitziation losses. And though he used an example with sRGB on the other page, which is a smallish 8bit color space I am not sure he referred to 8bit sRGB to 8bit Lab conversion. He is not specific enough for me to understand him well.
Anyways, your test and numbers are convincing.
C1 Color shift with different zoom levels
The viewer in C1 has its own zoom level on Windows too.
You are right, the other raw developers do it better, there is no color shift with small zoom levels or small output resolutions, but your example is an edge case, as you said "pushed very far", it is heavily underexposed as I can see now after downloading your file, and not even base ISO.
Still, you made a point, color is judged more accurately at 100%.
Maybe this is an argument for 5k monitors when judging overall color harmony of the image as a whole because the zoom level is closer to 100% when zoomed to fit (but not the fastest with C1 :-)).
Best regards
BeOEDIT: I used the color pickers in C1 to show the underexposure, not the color shift between zoom levels.
You used the color picker in PS, which is more reliable as you used it on the exported images. PS pickers have a bigger patch size, right? I am highly dissatisfied with the C1 WB picker, patch size is too small, maybe even only one pixel.!?
0
Post is closed for comments.
Comments
19 comments