Ever since i upgraded from NLP 2.1 to 2.2 i am getting horrible conversion results. I tried different scanners, different scanner software, uninstalled and installed multiple times but something seems to be very far of when im doing a conversion within Lightroom Classic (Adobe CC Version). I have experimented with a lot of different settings. The colours are way off and heavy vignetting in the corners. I am currently using a Epson V600 with Vuescan but also tried a Hasselblad X5 and having the same issues. When i convert normally within vuescan everything seems well exposed and balanced. Can somebody maybe notice something about what i am doing wrong? (Shooting Portra 400 btw 1 stop over)
Thank you so much in advance, this is driving me crazy.
I noticed, mine went straight to spam, maybe check that? If not here are a few more different download links of the same files: (sorry not that experienced with file hosting websites)
Left:
Converted with NLP, Basic, Pres-Sat: 2, Cinematic Rich, Neutral, Crystal (100), tight crop
Post conversion: selective sky darken
Right:
Above: Converted manually, NLP would not give me suitable output.
Below: Lightroom did not show a proper profile, so I selected the profile manually. After this intervention, and raised exposure to spread the histogram, NLP produced the following without too much fiddling:
For images like these, I propose to use a fairly tight crop. You can always adjust it later.
You might also want to check that Lr reports a proper profile for the images.
I notice a fair amount of dust. Better clean the negatives before scanning. Dust can be removed in post, but it also influences the conversion. A clean negative helps to get better conversions.
Regarding “weird”: the shots look like they were taken a long time ago. Colour film used to be less “good” at that time and they seem to be less stable. Over time, colours can shift and some of it makes for that vintage charm. Don’t expect “digital” quality, unless you’re ready to do a lot of post processing, which can be done if you also save the converted images as tiffs.
Thank you for trying these images with your install of NLP. So you kinda got the same colors as i did with a standard initial conversion? That means there is nothing wrong with my NLP install? Is there any way to check? The dust issue is a good suggestion, i will try to keep them a little more clean. What do you mean with converted manually btw?
The funny thing is that these shots were taken in August and developed a month ago, this is fresh Porta 400, always stored in the fridge.
Your installation seems to be okay. One thing to remember is to UN-CONVERT when you relaunch NLP on an image that it has already processed. But we don’t have this issue here and now.
As for your images, I notice that Lightroom shows me histograms that look underexposed by 2-3 stops, while RawDigger shows histograms with a cut-off red channel, but without signaling over-exposure. Strange, I’ve never seen this with my camera-scanned negatives…
Manual conversion:
Look at the histogram the NLP produces: R, G and B tone curves are inverted and shifted around (this is a blunt description of what goes on behind the scenes). I did just that… (roughly)
Okay that is one thing less to worry about. I am aware of the UN-CONVERT need. The fact that lightroom says my images are underexposed is strange. The cut off Red channel is even more strange. I can see the cut off red channel appear in my images, some spots that should be red just appear as a black dot with a tiny line of red around it.
So it might be my scanner then? Or the scan software that i’m using? I’m switching back and forth between silverfast and vuescan to which is best but both are not really doing that well. I see a well exposed image when i preview it as a positive in either software.
Notice that in my histogram after NLP conversion there is a dotted reddish line in the left part of the histogram. That might be the cut off red channel.
One last thing i noticed is that my images out of the software straight on my desktop look very different from the ones i see after i imported them in lightroom. Maybe that has something to do with it?
Lightroom sees DNG files as raw files and applies whatever setting Lr intends to apply. This can change the aspect of colours considerably, specially if they are nearly saturated. One way to see the differences is to export in the same format and with the same colour space.
You could try setting your scanner software to create an sRGB or AdobeRGB TIFF instead of a DNG and see what the differences will be. Most apps apply some “ingredients” to an image and none adds the same… When an image is first imported in Lightroom, Lr displays the preview that is often built into a file. Then the display changes and the image looks completely different. This is most obvious when the (digital) camera is set to B/W, which will add a B/W preview to the colour raw image data.
I set Lightroom to apply a preset that is fairly neutral and corrects CA. I want to add bells and whistles myself and first start with something plain and flat. I often also use the NLP profiles for that
I have strange results with version 2.2 too. It was much easily to get good colors and color balance with 2.1.2, i had to downgrade to it. But i working with dslr scanning.
The results from 2.1.2 look so much better for me out of the box, and I can’t make the 2.2 results look even close to it. I’ve spent so much time trying to figure out what was going on before I realised it happened after the NLP update.
Have a look at the comparison below.
The one on the right is converted using 2.1.2 - the colours look so much more vivid and there is a decent amount of contrast.
The 2.2 version on the left looks quite flat and the colours are nowhere near as saturated. In fact a lot of my shots look like they’ve been overexposed by 3 stops. When I pull the brightness slider to -30 the colours and saturation come back to what I would expect (and get if I scan it as a negative in SF). But if I pull the brightness down so far, the highlights look clipped and the picture too dark.
It’s not the biggest issue in this picture, but I have other examples of pictures I really like which I would’ve put straight to the bin with the 2.2 conversion.
Both of these are from the same 48bit HDR RAW dng from Silverfast).
I’ll link the raw files as well.
Any tips how to solve this? I will just stick with 2.1.2 for now and actually re-process all of my recent work with it because I was struggling so much trying to figure out what was going on, even testing my camera, light meter and scanner.
One thing I notice looking at the original negatives is that it appears that the film was pretty heavily overexposed during original capture. (Or perhaps improperly developed).
Combined with having a low dynamic range scene, this leaves very very little information for Negative Lab Pro to work with.
This will make the results more fickle and harder to adjust. Because there are fewer points within that range to work with, rounding errors will also increase. NLP v2.2.1 may also be trying to extract more information from this little data than it should (v2.1 could handle this scenario a little better since it was not trying to extract as much info - version 2.3 will do a better job in this scenario).
A little “hack” you can do that may help in this scenario…
Before conversion, use the main RGB curve to give yourself more leeway… with the main RGB curve selected, move the upper-right point directly to the left, just up to the point before your negative data starts to clip (it’s OK if the border clips).
No problem at all. Thank you for your solution, yes these very dense negatives might be pushing the boundaries of the scanner and software. I am definitely going to try to expand the RGB channel a bit!
I really love the results that this program is giving me on the 99% of my images. Great to hear you are still improving and mentioning a v2.3. I am very looking forward and curious about the improvements you made.
Thank you for stopping by and taking a look at my results. Appreciate it.
