My grandfather shot a lot of Agfachrome back in the 60s and 70s, and I think even into the early 80s.
The slides themselves weren’t really stored with the most care, so they’ve turned purple over the years. I saw someone who managed to invert the tone curves to create a negative, and then convert with NLP with decent results on correcting the colors, but this really seems to be dependent on just how bad the slide itself is as my attempts at is resulted in a positive that still had somewhat wonky colors (in fact, even closer to purple!)
I’m wondering if there is an easier way, or if I just need to buckle down and play with the tone curves by hand to get the most desirable results.
Please note that NLP provides a starting point, and, according to IT rules, we get GIGO (garbage in garbage out) results. Old slides need some extra effort and they deserve it - if you think them worthy.
I’ll give that a shot after I scan some more; some I know for a fact I will not be scanning, so I won’t waste my time on them. It’s mainly the Agfachrome slides that seem to have developed this purpling. I have Kodak slides from the same era and they look great.
After color correcting a photo of my dad from 1967, I’ve come to realize that for the best results I’m going to need to make use of masks as well to avoid blowing out or crushing other parts of the exposure when making adjustments. Again, I’ll save this effort for those shots that are worth that kind of effort. Many of these I’d like to actually print as we’ve never had prints of many of these.
Thanks for the input! Should I come up with a reasonable workflow that is repeatable, I will be sure to share my findings.
Interesting. I have a bunch of Agfachrome taken in the 70’s, that have pretty much all shifted towards purple/blue. As mentioned, I did a simple tone curve invert and sent it through NLP, with great results. The results from the initial conversion, hardly required any additional processing.
I tried this exact step with two shots and didn’t get the results I was hoping for; perhaps I messed something up! You’ve demonstrated it does indeed work, so I’ve no doubts about that.
This evening I’ll grab a carton of Agfa slides and scan them in and try again.
Will do! I ended up taking a nap last night that instead of being about half an hour ended up being closer to four, so I went to bed. Been a long week. My plan was also to scan in my roll of Phoenix 200 I shot at 125.
Had the same problem here with slides from the 1950’s. I scanned and converted them before I knew about NLP (…), so did everything by hand, which was incredibly time consuming and tedious. What did help significantly in getting the colors anywhere close to what they should be, was to set the black and white points in each of the R, G, and B channels individually in the tone curve panel. This got rid of most of the color casts (although the scans remained rather crummy). Perhaps you can try to do the above step first, then flip the luminosity curve to create a negative, and then run it through NLP?
I’ve been meaning to try this, but the last few days have been absolutely hectic with work. I’ve been getting home at 2300 every night and have had to work the weekend.
Tonight I will try to sit down if I get home early enough and have the energy to do so and share my results.
I just noticed your post - sorry if this response is too late to be relevant.
Based on my experience you will “just need to buckle down and play with the tone curves by hand”!
I have scanned many hundreds of faded Agfachrome transparencies with generally good results. However, I found that the only reliable way to get good colour is to manipulate the R, G and B tone curves individually (and often quite drastically, especially the blue). I found Photoshop best for this because after I get reasonable colour, I can use a curves layer set to luminosity blending mode to adjust the overall tone without introducing unwanted colour shifts. I then use Lightroom for final adjustments, and I’m looking forward to NLP v3.11 with adjustments for positives!
Agfachromes are among the worst for fading, which is a shame, because I really liked their colour rendition when they were fresh, back in the 1960s and 1970s! What makes it worse is there is typically a large variation in fading between slides from the same box, in addition to box-to-box variation. This means you’ll probably need to process each slide individually.
When you say, “they’ve turned purple”, it means that the dyes have faded, and information has been lost forever. The best you can do is use what remains to create an impression of what the slide had looked like originally. In principle it’s the same way that NLP allows you to interpret a negative to give your preferred impression of the original scene.
In general, I treat all scans of old slides as ‘reversal negatives’ that need to be interpreted. If they aren’t Kodachromes, they’ve probably faded, and if they’re Kodachromes they probably have a lot of usable detail hidden in the shadows.
Original slide (from about 1955) and two versions done with NLP.
The film was manually sandwiched between to fairly robust glasses held together by the sticky masks and two tapes along the top and bottom sides. The sandwich was never opened and I’m not sure if it is Agfachrome (or whatever the product was called then, probably Gevaert).
Nevertheless, NLP did a fairly good job to begin with. All I added were some tonality tweaks and a pinch of structure.
Still relevant! I’ve not been able to sit down and try again as I’ve been using my camera more for work and streaming, so I’ve not been setting it up on the scanning setup to try out a few more results.
It seems to me, as has been mentioned by everyone here, that NLP will give a good baseline from which to work and that all of these older Agfachrome slides will require some love. That said, as someone also mentioned, I also agree that this work involved should be reserved for the Agfa slides worth doing it to. I cannot see myself doing every single slide at this point as I’d be here for the next 6 months at least!
I’ll have the next two weeks off after this coming week, so perhaps I’ll sit down and try again.
I’ve spent some time revisiting what I’d done more than 10 years ago. Eventually I recalled that I’d used VueScan ‘Restore Fading’ and its Histogram controls (clipping and tone curve) to get a starting point for final adjustments (with Photoshop in those days). I had been working with ‘actual’ scans, but the latest version of VueScan can read raw digital camera files, so digital camera ‘scans’ can be processed in VueScan, with access to its Restore Fading and Histograms features.
I also confirmed that I could do everything in Photoshop without using VueScan, but it takes far too much time to be practical for more than just a few images.
Getting the VueScan tone adjustment roughly “right” is worthwhile because then you can use Lightroom for final adjustments with the R, G and B tone curves and just minimal use of composite RGB tone. The problem with LR (and ACR) is that the RGB tone adjustment isn’t independent of the individual R, G and B tones (with PS you can make the RGB tone adjustment independent of colour by setting it the luminosity blending mode).
Anyhow, it was interesting to go back to what I’d done and to see what I could have done better.
I think before I dive back further into this, I need to solve my vignetting issue when scanning. I’ve also been using my camera for work and family events, so it’s been off the scanning setup.
We’re expecting some crazy weather in a few days, so work will likely give us the day off as we are ill-equipped for Wintery-mix precipitation.
There’s a popular thread on vignetting that I need to read through again but every time I get home from work I get distracted by a million other things.
