No, I don’t think my adjustments went far enough. But I do think further messing with red and blue channels would eventually make things look right.
Well, my votes are no, and no. My findings indicate there’s a white balance problem. If there was a ‘decode’ problem, it wouldn’t just limit itself to a certain color. And I’ve shown that the camconst.json numbers will provide a well-balanced color rendition, see post #15.
Expressing white balance as temperature and tint is a software abstraction, attempting to accommodate white balance in the terms you think of it when you shoot the scene: the actual Kelvin temperature of the light. After the camera records the light, that notion goes out the window and is now a matter of how a white thing in the picture can be made to look white, that is, its R=G=B=maxiumum_value. I personally find temp/tint to be confusing in post-processing for this reason. So, I start with considering the R, G, and B multipliers that the camera provides in the metadata, which is also how the image in post #15 is rendered. In that regard, your raw file seems to have a decent set of numbers.
Given the RT camera primaries and white balance multipliers ‘work right out of the box’, my surmise is RT is doing something of its own, likely with white balance.
This comparison would make sense if you used shots of a color target, not some random scenes with uncontrolled lighting. Currently it’s “I like Lipton” vs “I like Earl Grey”.
I understand your point, but what I photograph are “random scenes with uncontrolled lighting”. I used two of them as examples where CaptureOne seemed to get it right and RT did not, even with extensive manipulation. I didn’t cherry pick them either… RT seems to reproduce most if not all my shots with a yellow tint that is difficult to fully eliminate. CaptureOne seems to reproduce exactly what I saw in the viewfinder.
So why not just use CaptureOne? Well… I am using it right now. But their upgrade pricing from v12 to v20 made me mad and I’m trying a deep dive into RT in the event I decide to jump off the CaptureOne train.
Ok, having more fun here in the loft; in RT I went ahead and made better settings for the RGB curves, added a tone curve to get that contrast. My first one was a bit yellow, compared to this one; is this where you want it to be?
OK, had a look on my main monitor. It looks close, but too cold. It is lacking the warmth of the original… to me. However, it is certainly better than I was able to get in RT. Would what you did be applicable globally or was it specific to this one image?
I have lots of similar images. I discovered that a local undeveloped park has a LOT of mantises and I had fun last summer shooting them. They actually pose when they are trying to run away. Next summer I’ll do even better now that I have a decent macro lens.
The rawproc image was rendered “as-is”, all I adjusted was tone. I kinda like it… if you want it a bit warmer, small adjustment of Temperature should do it. For the RT image, I was trying hard to avoid “yellow”, maybe went too far. Again, I think the Temperature slider in small amounts would take it where you want to go.
Oh, I didn’t explain, rawproc is my own software, but it uses RT color information.
I re-looked at your original images, and there’s a lot of processing going on in each. That confounds your assessment of what to do next; at least, that’s what was happening to me about three years ago when I restarted photography. Here’s what I see, and recommend:
RT’s camera color primaries for your camera looks fine to me. These don’t usually mess with color; they’re built to encompass all the color your camera can capture, so it can be eventually downsized to color gamuts suitable for display.
White Balance is where you need to spend a bit of time considering. The first picture has mixed light, and really neither the outside or inside is properly balanced. I think a lot of the yellow in that one is the warm interior light, not balanced in the camera. You’ll only be able to accommodate one light source in such situations without doing a lot of local editing. Your first RT mantis image looks yellow, but again, there’s a lot of other processing going on in that image. I originally thought otherwise, but on second review I didn’t see any yellowing in RT that looked out of place in the Neutral profile starting point. Good thing about shooting mantises in the park, it’s all daylight. When you go out the first time, bring a white or gray object and include it in a photograph. RT will let you select that object to set the white balance, and I think you’ll find that setting to work all summer, in most daylight.
After color is figured out, adjust the tone. This is the contrast, dark/lightness, etc. Now, being too aggressive in tone can change your color, so use a light hand with the sliders, curves, and other tone tools.
Now, for RT, I recommend this: In the Preferences dialog, under Image Processing, set the Default Processing Profiles to Neutral. Now, every time you open a raw in RT, it’ll look dull, but the colors should be discernable enough to have a decent idea of their base look. if not, try hard to work it with White Balance; that is the first place you’ll encounter discrepancies. After that, for an easy end, try the Auto-Matched Tone Curve; this tool takes the embedded JPEG and attempts to approximate its tone. But eventually I’d suggest you try making your own curves, that’s where you’ll get the most control over tone in your images. The mantis renditions I posted required rather aggressive curves to get the contrast. RT has a rich set of tone tools, maybe a confusingly rich set, but you can do most anything you need to your images, tone-wise. if you’re adjusting a lot of things and can’t seem to control it, just reset the profile to Neutral and start over; it’s not really a great setback to do so.
Disclaimer: I’m a very intermittent RT user, so if someone pipes up and says Glenn is full of it, take them seriously…
The bottom line: start Neutral, work color first if needed, then tone. I think this is a good approach for any software, really.
Finally, I know Capture One made it easy, but it does that at the expense of not showing you the “why” of what works. You’re a bit in “manual transmission” territory now, and RT combined with this forum is a good place to master it.
Ah, just noticed, picking around the RT tools, the default camera primaries come from a custom DCP profile, “SONY ILCE-6000.dcp”, so they’re not the same primaries as I’ve been using. This is under “Color → Color Management → Auto-matched camera profile”. When I select “Camera standard” instead, the colors get a bit less yellow, and align with what I’ve been starting with in rawproc. This might be the source of your angst.
I loaded your pp3. Initial impression was the color wasn’t too bad. But the image was dull and as soon as I moved the black point to lower the blacks, it revealed the same old yellow cast. So no real improvement.
Changing from “auto matched camera profile” to “Camera Standard” does remove a bit of the yellow. But it also seems to remove some red, leaving the overall image too cold and washed out. If I play with the saturation and color temperature I can sorta correct that, but the yellow cast is back.
Setting the profile to “natural”, clicking “Auto Levels” and making my own curve gets me to a decent starting point. Thanks for that suggestion. Tweaking the bottom end of the curve gets me very close. In fact if I didn’t have my CaptureOne result to look at I might call it a day. But there is still a yellow cast that, while minimal now, won’t go away altogether no matter what I do.
UPDATE: After processing per the above and moving the “tint” slider left a bit, the yellow cast seems to be (mostly) corrected. Hooray!
A) I should not have to struggle this hard?
B) Why does “Auto Levels” leave the black point such that I have to tweak the bottom end of the curve? Does it merely adjust the exposure so the brightest pixel = 255 and leaves the black alone? Its a little unclear how it works and exactly what the histogram is showing.
You have a point, sir! And for the most part, I enjoy the challenge. To a point…
Actually, I was referring to the prospect of having to do this kind of careful tweaking with every shot. I would prefer not to have to pay this much attention to color balance since the camera does a pretty good job most of the time and CaptureOne translates this to the final output without intervention 95% of the time.
The only thing i can think of is that it looks for some sort of threshold of number of pixels to set black, and there are enough to trigger that even though one would rather put it higher in the histogram. A RT developer may have to weigh in.
Okay, post your .pp3 for this minimal yellow. I have a feeling it’s a matter of interpretation at this point, but won’t know until I see your raw processed with what is producing the cast.
The struggle lessens as you learn what’s going on under the hood. This is the place to do that…
Once you find the ‘sauce’, it should give you decent results for any image exposed under similar conditions. 'Specially if your scene is illuminated in daylight. i still would pay close attention to white balance, shoot an image with a white patch or such. Your camera may have a measurement tool for the white balance presets, where you can aim at a white or neutral target, press the button, and the camera will calculate the white balance multipliers and store them in the preset. My new camera has that, and it works peachy-keen…
I’ve worked hard to find such a workflow, and I’ve almost arrived. Took three years, but it revolves around a parametric tone curve called filmic, other threads here can give you the history of oil on it. For color, I find the simple primaries from the dcraw and camconst.json files do the trick nicely. And, I measure white balance if in anything but strong daylight. Still though, I find most images require a bit of specific attention, as I also am shooting in a variety of lighting, shot-to-shot. That’s just the way it is, with any most any software.
I do hope we’re scratching your itch here. Really, there is no better place to learn this stuff than here; professional-grade free software, and the developers hang out here and will directly answer your questions. However, I think I’ve short-circuited that in our discourse; they’re probably sitting back, watching me flail with RT…
Um, I could have sworn his EXIF said it was a 6500, not a 6000… Where are you getting a 6000 from?
I’ll need to check again when I get home. Maybe I ran exiftool against the wrong file. OP has never explicitly stated what camera he’s using, but didn’t disagree with me when I said A6500…
If somehow something is assigning A6000 color matrices or profiles to A6500 color matrices, somehting is likely to go very wrong, because the A6300 introduced an entirely new sensor. A6300,6500,6400,6100,6600 in theory use the same sensor but MIGHT have different CFAs.
(why that oddball ordering? That’s chronological order of release of the bodies in question.)
Edit: Wait, the two examples the OP provided are coming from two completely different cameras??? One is from an A6000, one from A6500.
Would have been nice if the OP had mentioned these are examples from two different cameras.
For the indoor shot, artificial lighting combined with a single-illuminant color matrix might be a contributor. For the outdoor mantis A6000 shot - not sure.
(RT has a dual-illuminant profile for the A6000, but for A6500 just has the basic color matrix)
Yes, the two images are from different cameras. The mantis was shot on my a6000 and the room was shot as a test with my a6500. However, RT has a profile for both the a6000 and the a6500 and the yellow cast is the same in images from either camera.
I have managed to get close to what I would call accurate color rendition using the neutral profile and with some manipulation of the color temperature sliders and/or the green sliders in the RAW tab but nothing so far has been what I would call “spot on”. The same files processed with CaptureOne with NO color manipulation always look much closer to “remembered reality”. Whatever the issue is, it does not seem to be a simple RGB levels issue because that would be easily addressed with the channel mixers? It “feels” more like a gamma or curve issue with one or more individual colors.
I have been playing A LOT with RT on many images and every time I think I’ve got the color to what I would call acceptable I flip over and view the image in CaptureOne and find I’m not there yet. The yellow bias is still there in varying amounts.
BTW, I have found that the auto levels control is indeed working as intended. By enabling the over/under warnings, I find that there ARE outlier pixels at each end of the range (more than I expected) that cause the levels to seem to be compressed.
Yes, the lighting in both examples is not simple. There is a small amount of daylight in the high ISO room image and the light in the mantis shot is likely affected by reflections from all the vegetation. However, it did not seem to materially affect the image in the camera viewfinder or what CaptureOne shows. So I’m not sure it is the culprit here.
And for the most part, I enjoy the challenge. To a point…