Recent NLP conversions have not been successful

[klydo]

Hi y’all - I bought NLP only a few weeks ago after running the trial session. I was really satisfied with the results compared to ColorPerfect and continued to be up until this past week. Current setup is an Epson V700 with Vuescan.

I developed a bunch of film this past weekend, but each scan I’ve run through NLP (either as a TIF or DNG) gets completely mucked up. Insane cyan or green cast, highlights blown out, etc… I’ve tried resetting my Lightroom preferences, deleted & re-downloaded NLP, triple checked I haven’t accidentally added any adjustments on Vuescan, made sure the WB was set to 6500 (also tried sampling the film base), everything I could think of.

I ran the same exact TIF negative file through both ColorPerfect and NLP, and these are the results, without any further tweaking on my part (top is from NLP/Lightroom, bottom is CP/Photoshop)

As I mentioned, NLP was initially working great for me until this week. I did just download NLP’s “Sample Raw dslr scans” folder and imported them to Lightroom, and the conversions for those photos were perfect. I’m not sure what’s going on with my own DNG and TIF scans.

Few other samples of both DNG & TIF (various film stocks)scans after running through NLP:

I have a gnawing feeling that this is some sort of user error on my part, but for the life of me I can’t figure out how to fix it. If @nate or anyone can offer any guidance or suggestions, it would be greatly appreciated. Thanks!

[nate]

Hi!

We’ll figure this out for ya!

There are a couple of things that could be going on here… the easiest way to figure it out would be if you could PM me or email me a few samples that I can try to convert on my machine. (My email is nate@natephotographic.com, or you should be able to click on my username to private message me).

I will say that I have seen some issues with v700/750/800/850 in Vuescan DNGs in particular, related to an issue with the embedded profile. There will be an updated profile in v2.1.

A number of users have also gotten better results using the EpsonScan method noted here, so that may be worth a try.

If you have found that Negative Lab Pro worked well with previous scenes and suddenly seems to not work well with these scenes, it could also be that the scenes themselves are trickier for Negative Lab Pro to interpret… the initial conversion are just a starting point and will often require some adjustments (though usually not as extreme as the first example).

-Nate

[klydo]

Hi Nate - just sent you an email with some TIF & DNG files! Thanks!

[wally]

Bump on this one.

Having a similar issue. However, I’m scanning with DSLR RAW.

It was working fine a number of weeks ago. I even pulled an old negative scan that I know converted successfully before. However, it had the same issue. I’m following the same workflow as instructed on this site.

Below is a screenshot of what it looks like right after the conversion:

[nate]

Hi @wally - for the above image, you will need to crop out the sprockets prior to conversion… (alternatively, you could increase the “border buffer” amount, but in this situation, you will get more certainty by cropping manual prior to conversion - you can then un-crop afterwards to show off the sprocket holes.

Why? The sprockets are letting in direct light from your light table, and since you’ve left them in the crop, the “black point” of the conversion is being set to the light table light, rather than the film itself. That’s why the film base (which should appear black) instead appears to be a bluish grey.

-Nate

[wally]

Nate, thank you for the reply and apologies for posting this in an unrelated thread now that I know what the problem is.

It’s funny, now that you mention it, I remember when watching earlier videos that you’re supposed to crop out the sprockets before anything else…

Don’t know why I forgot, but it’s been a while since I’ve used NLP.

Thank you so much for enlightening me! Apologies again for the facepalm!

[klydo]

Hi Nate- just wanted to check to see if you got my negative examples via email? If not, happy to send them via this forum. Thanks!

[nate]

I did! I was away for the weekend, but looking at it right now!

[klydo]

No worries! I’ve had some issues with my email lately so just wanted to confirm, thanks!!

[nate]

Ok, just looking at a few of these now…

First, there is a known issue with some vuescan profiles in NLP v2.0 (including v700/v750) that can cause issues in some scenarios… new profiles will be released in v2.1, but in the meantime, you can get profiles here: https://www.dropbox.com/sh/76flmem0qsootc6/AACjiNNSaH-QaTTNu7xzl4LTa?dl=0

The bigger issue is that these are definitely tough scenes…

Anytime you have a scene with two very different lighting scenarios in a single scene, it makes the conversion more difficult.

For instance, in this photo, your subject is in the shade and the background scene is in direct sunlight.

We can improve the color balance here simply by turning on “AutoColor - Warming” (note that I have also set the “Tones” dropdown to “linear” to provide a more neutral tonal starting place for editing).

Now the color balance is improved, but the skin tones still feel very dark and drab (because Negative Lab Pro is balancing for the sunlit background in the scene as well). We can fix that by pushing up the brightness a bit and tweaking a few other settings:

That doesn’t look bad. Maybe still a bit flat and not the healthiest looking skin tones, but this is probably the best we can do while keeping details in the background.

If you don’t care about losing detail in the background, you can crop in to the area in the foreground (before you get full on direct light in the background), and make a conversion cropped like that. Here is what you would end up with:

The skin tone looks better here (in my opinion), because the entire evaluation is done for the lighting conditions of the subject - but of course you lose the background.

For the TIFF you sent, the biggest issue is that it needs to be converted from Gamma 1 to Gamma 2.2. prior to using Negative Lab Pro. There is an included utility for doing this, that you can access in Lightroom by going to “File > Plugin-Extras > Tiff Scan Prep”.

After doing the tiff scan prep, you can then convert the new file it creates using Negative Lab Pro.

It will still require some color and tone balancing to get the look that you want. Here are the settings that would get you close to example you shared:

Hope that helps, and thank you for sending file

-Nate

[klydo]

Nate- thank you for taking the time to look into this! This is a obviously a huge help. Excited to get back into scanning the rest of my negatives with the new profiles.

Really appreciate the explanations as I would never have figured out a way to solve these issues myself, and it’s interesting to see a bit of how NLP works. The issue stemming from the extreme lighting differences makes sense.

Just out of curiosity regarding the TIFF file as a Gamma 1 to Gamma 2.2 - I’m assume this was due to however I had set the Vuescan output settings? Is there a way I can ensure it’s being saved as a 2.2 moving forward? I’m seeing a bit on Vuescan’s support website that it has to do with the chosen color space, but is that the only setting that would affect the Gamma correction?

[nate]

Were the tiffs made as so-called “RAW Tiffs”? For instance, in the output module of Vuescan, if you select “RAW File” but do not select “RAW DNG Format”, it will output as a TIFF file with very neutral settings and gamma 1.0. In this case, the “color” module is not applied to the output, and I believe that applies to color space as well. So… I don’t believe there is a way in this scenario to output as gamma 2.2.

If you are saving directly as Tiff (i.e. you just select TIFF in output, and not RAW), then the color tab will have an impact, and the output will have the same gamma as the color space… so for instance, if you set it to Adobe RGB as the output space, it will be gamma 2.2. BUT this could also create other issues, as Vuescan essentially attempts to manipulate this as a regular image.

[klydo]

They 100% were saved as RAW Tiffs. I had no idea that had the effect it did. Thanks for clearing that up!