Bad results after upgrade to 2.2

[unfon]

I`m and my friend have problems with color on NLP 2.2 too, but with DSLR scanning workflow.

Maybe we did shootings in trouble conditions, but all the time no one image is automatically converted with good result. We are shooting portraits and its very important for us to get the right or neutral colors, because we have the skin tone. Each Image we need to adjust manually - and light\contrast and color. Its not simple at first, but i have practicing and understood how to working all the sliders to get the best results.

With 2.1.2 we can setup really good, natural, saturated, color variated image. But after upgrade to 2.2, i cant repeat my results on images, that i already did with great result on 2.1.2 or below. Maybe with new images its not too much noticeable, but with old images Im already know, how they should to be and i cant to make good results. It may not be clear what I’m talking about without examples, but the same problem with each image. If with 2.1.2 after manually corrections i have neutral objects is neutral on image and at the same time i have truly colors in shadows\midtones\highlights, with 2.2 i cant to get the same colors. Or neutral objects are not neutral, or they are neutral, but all other with other tint. Or i get the neutral result, but color is not rich or saturated. I tried all the ways to work with color temperature tool, shadows\midtones\highlights corrections, but the best method was to downgrade to 2.1.2. Im not working with custom luts, styles, nuritsu\frontier style, just the basic converstion with default saturation, all camera profiles applies as usual.

Maybe version 2.2 theoretically, mathematically is much better, but on practice is not for me. It would be good to add “the legacy style conversion method” as an alternative in future versions or fix this. I`m saw some topics with similar problems, but, anybode, do you have problems with the color after the update or am I the only one have?

[Digitizer]

Please help us help you by posting screenshots of an image and its conversions done in 2.2 and in 2.1.2 or earlier. Providing an original raw file can help too.

[nate]

@unfon

It would be good to add “the legacy style conversion method” as an alternative in future versions or fix this.

Yes, I’m working on something like “process versions” - where you choose which engine you want to use, or even make adjustments to the engine itself.

Still, it should be fairly simple to get the same or much much better results using v2.2. If you want to post a link to a sample RAW that is giving you trouble (or email it to me at nate@natephotographic.com), I’d be happy to take a look. It’s possible that something else has gone wrong, or there may be something about the file that is giving v2.2 trouble.

Thanks!
-Nate

[unfon]

Thanks for answer Nate, sorry for the late reply)

It would be really nice if we could choose the process version, because I really don’t like the new one and I usually have to replace the folders with 2.1.2 and 2.2 versions. I have the photos that I already done in the previous version and if I want to modify the settings of conversion, I need to open it in that version. A few times I opened old photos by mistake and clicked save settings… no way back, I need to convert again.

For me, all the sliders worked in 2.1.2 is preferable. A little bit different in contrast, but it`s maybe a not noticeable difference. This is a strange decision to add sliders of color temperature, that controls blue and green channels separately, we already had those sliders in the mid tab. The new sliders are not the same, but solving the same task. Theoretically, color temp slider should be the combination of two or three channels. In 2.1.2 the main part of the work in mid tab, shadows and highlights mostly + - 5 maximum, in 2.2 i near the maximum in the mid and not nearly the final result.

I noticed that the algorithm works differently on separate images, and maybe more accurate on more flat images, without extreme highlights, shadows and with more neutral colors. I attached the photos, I’ll try to explain what I don’t like. Everything was shooted by me with a kaiser table, no filters, all my conversions were nearly the natural look, like in really. I tried to repeat the result in the new version. Identical crop and color temp of the raws. Not the perfect work, but I tried to get the reference look, maybe not get the contrast, trouble with colors and color balancing.

1. 2.1.2. All looking good, may be more cold or warm, more clear colors.

2. Converted with 2.2. Maybe the difference is not noticeable, but I can’t get previous colors, skin tone is more dirty, on the middle of the shoulder for example. And this problem with all images, with the old version I can get more clear colors, with new a lot of unsaturated colors in the shadows or skin tones.

3. Comparison of crops of the 1 and 2 images. A little bit difference with color temp, but if I move the color balance like in the 1 reference, all image looking worse. You can see here more trash colors in skin tone, under the right eye and cheek.

4. 2.1.2

5. It is a good illustration of the problem, if i opened the image from the previous version in new. The more the sliders move, the more the difference.

6. It is the example of image, that so far from the result in the starting conversion. Need to remember how it was looking in real life and need active work with all sliders. Need to set larger values in the channel tabs and the small limit does not allow it. My luck, I was nearly the reference.

7-8) 2.1.2. The result of the conversion of a small spot of the image, i have lab scan, it`s similar. But i don’t know how to repeat with similar results.

I want to get just a natural image with good colors and after do my own color correction. But it takes a lot of time to get the first result, a lot of attempts with different color temp, playing with crop, copy from previous shot. if I will find more problematic images, I can show them later. If the image after first conversion is 95% near the result, maybe its not a problem, just a little move the contrast, a little color balance and its done. But often this is not the case, missing exposure, contrast, colors, color balance, all is shit and in the new version in more harder to work with this images, or just impossible.

Link to the raw files on file sharing service.
Raw files

[nate]

Thanks for the examples and the feedback, @unfon.

Yes, I can see what you mean. Yes, in some circumstances, v2.2. is basically trying to add too many points to Lightroom’s tone curve, and it is creating the sort of “posterization” effect you see here. If you can hang on for just a bit longer, this will be 1000% better in v2.3.

Quick look at these RAWs processed in v2.3 (releasing soon):

And then, using “Sync Scene” from this to the other shot of this model:

I should also note that if you want that super “soft, creamy” look, that is easy to do with the tonal adjustments…

• Start with “linear” tone profile.
• Adjust the WhiteClip to ensure no important highlights are blown
• Then, for the creamy look, pull down the “Whites” slider, and push up the “Lights” slider.
• Adjust the BlackClip to add more punch, then push up the “Black” slider to add soft fade.
• Adjust color balance to taste.
• After applying Negative Lab Pro, adjust the “Noise Reduction” in Lightroom to soften the skin further.

Hope that helps!
-Nate

[unfon]

Yeah, you are right, maybe it looks like a pasteurization effect, or maybe a desaturated spots on some photos.

v2.3 looking maybe better, but your conversion is more monotone and undersaturated. I think you used another Pre-Saturation, maybe 3, I prefer the maximum (5), it`s not desaturating the raw file, getting more natural colors. Low values are decreasing the color diversity. So I can’t say for sure it’s looking better or not)

Im always using a Linear curve, in all cases, its getting the best result, more freedom working with contrast. And overall, I haven’t got problems with it, just adjusting the brightness, shadows, highlights, and v2.1.2 looking better today.

But there are always problems with colors, color balance… This is a great success if everything is fine after the first conversion, or you can copy the settings from the next frame, but what if its impossible? I have images that i cant process, and I understand that the problem is not in NLP, dslr digitizing gets the best result. I have a v850 scanner and it cant too, similar poor result with any software like a VueScan, Silverfast, Epson Scan. Its a very good, if you remember what exposure you set, what color balance was, what color of clothes in real life and manually trying to fix this. What can you recommend in this case?

For example, I have this image that I can’t process with natural colors, like I did before. Previous photos were with Portra 400, this is Gold 200. And i can’t save the natural skin tone, red dress and overall natural color balance. First image is my try in NLP, not good colors, especially the color of red dress. In real life it was a clear saturated red color, like a R:255, G:0, B:0. But in mine scan its more cyan. Im already warm the image, plants lost their greenish color, dress is still mismatch. Maybe its a too much red for scanning, red color just clipping and i need to reshoot with 80a conversion filter? I dont know… I bought a filter recently, but not tested it.

Second from a local popular lab, they got the natural colors, but much oversaturated, dress clipping again and has not detail)
RAW File