My SIGMA telezoom is unfortunately not covered by the lensfun database and I lack the equipment and opportunity to generate the raw files necessary for producing the calibration data. Therefore I have contacted SIGMA to ask whether they could provide the data for all of their lenses. Could be a good marketing point.
SIGMA have been very quick to respond very politely, that they likely won’t do that, but the support guy said he would pass on the request to their engineering team for consideration.
It may help, if other users of open source photography software could do something similar. Rather than individual lens owners photographing office building fronts for lens calibration data generation, the lens manufacturers could be encouraged to provide that data.
Well, not surprising Sigma didn’t react enthousiastically. It’s the reaction we generally see to such requests (in their defense: I’m not at all sure they have the data in a form you could use, and collecting and formatting the information costs engineer hours, for something that wouldn’t benefit all that many customers).
One thing that struck me was you asking for data for all of their lenses, where you likely only have a few. That’s a bit broad as a request from a private client, and may not help getting the data you want.
I don’t know about the opportunity, but you do have the equipment: camera + lens(es) for which you want the calibration data (if not, why do you want the data?)
From the rest of your post, I see that you know the procedure. It’s probably faster to use that procedure (and upload the results to lensfun?), than trying to convince Sigma to release the info…
Now, if you manage to convince Sigma to release the information, please do the same for Tamron, making it clear that Sigma cooperated. After that, you could try the other manufacturers.
I have only one SIGMA lens. The reason I asked for all lens data to be provided was to make the case that it would be in their interest as a marketing point: Buy any SIGMA lens and be sure your software recognises it.
Nikon cameras include image correction coefficients in EXIF, but they’re in a format not recognized by Darktable. If we had the algorithm for how these values work, a lot of lenses would be covered at once.
At least that’s what I’ve been told here some time ago. @rvietor are you familiar with these embedded coefficients by any chance? I think someone said they’re in an Adobe format or something…
Yes, I’ve seen that. The problem with a long telezoom is that I would need a building with straight lines (e.g. an office building) that I can photograph head on from different distances. There is no such location anywhere near where I live in the Croatian countryside.
Nope, no idea how or where they are stored (well, makernote probably, but that’s basically an opaque black box). And Sigma lenses might report those values to cameras able to handle them (or asking for them). Dt can use such coefficients, you’ll see that in the lens correction module where it can say “embedded metadata” under “correction method” (if such metadata are available). So the code to read and use them should be in darktable
Funny part is: I do get that “embedded metadata” for a Sigma lens on a fairly old Sony body, and it applies a correction. But I can also pick “lensfun database” and get a different correction. Given the subjects and the type of lens, I haven’t tried to figure out which is correct.
My heuristic is a comparison to in-camera corrections. At least for first-party lenses (maybe 3rd-party native lenses too) they should be the reference point for lensfun imo.
It’s the SIGMA 150-600/5,0-6,3 DG OS HSM [S], i.e. the Sports version. Lensfun only has data for the Contemporary version, which is a bit different, but apparently more popular/widespread.
I was able to get something out of Gemini about the formula for the embedded Nikon corrections. It seemed to work OK in my limited testing. Perhaps someone else could pick up that work.
My annoyance is that the Adobe DNG converter seems to strip away the lens correction metadata from my Sony ARW files, so I can’t convert them if I want lens correction for my Tamron 50-300mm, which is not in the Lensfun database yet. I might get around to doing my own data at some point.
Not a stupid question. I would rather not, but the a7R IV only has uncompressed RAW that produces ~120MB files, and compressed RAW that are about half the size. But I don’t want a lossy compressed file. As far as I know, converting to DNG with lossless compression will produce essentially the same quality file as the uncompressed RAW, but half the size.
So, my goal was to halve the size of my files while retaining all the original quality. But there are compromises. One is the loss of lens correction via metadata, the other is a slight crop applied, which seems strange to me, but I can live with it. Also, there are various Sony-specific things you can’t do with a DNG, such as combine pixel-shift RAWs.
I’m still deciding if it’s worth it to convert. I’m trialing the compressed RAWs right now, but I don’t like the idea of having such a great sensor and then throwing away some of the data.
.