Blue/Purple fringing is an abberation from not perfectly correcting blue wavelengths to focus at the same focal length. So for the blue end (let’s say 450nm and below) it kind of makes sense to have some kind of roll-off to “hide” non apo-chromatic corrections, or better the errors that stem from non-apo-chromatic corrections. But those broadband large amplitude changes…that’s almost a stop of difference between certain wavelengths within the S-cone band.
In principle the same thing should exist at long wavelengths too, but at least for modern cameras there is the IR-cut filter which is has a rather steep highpass characteristic (OD5+ within 10nm or so).
Could those roll-off’s come from a spot-size shift at the sensor position? Meaning that because of a non-apo design the spotsize and thus flux-per-area changes wrt to the reference intensity measurement?
I’d love to see the transmission spectrum from a modern lens. I naively assumed that their transmission would be rather flat throughout the visible wavelength range.
Isn’t that Thorium doped glass that turns visibly Yellow? Blue or UV light curing might mitigate this.
Fascinating! The pentax has a lot of red and blue abberations, compared to my russian Tair 133mm lens, which barely has any, so you may be on to something.
Here’s a modern high end lens, an olympus 12-40 2.8:
Confirms those takumars were flatter than modern lenses! I think modern lenses try to maximise transmission as weighted by the luminance function, not equally for all wavelengths, so the results are brighter lenses (from the sensor’s POV) and are neutral enough. Also blocking UV could be advantageous? That’s my assumptions.
@Claes Well yes of course I have an industar 61L/Z! I have been taking my camera spectral measurements through that lens so far (as it seems quite neutral and I was not using it often). Will formally measure all my lenses tomrrw.
Every optic that is not truly neutral then would “stack” its filtration effect. This logic only works for cameras because you usually only attach one lens system. Attaching two lens systems (think lens+diopter or lens+anamorphic or lens+speedbooster) already would get you IMHO into colorimetric trouble. At least somewhat in the blue and red edges of the visible spectrum.
Now isn’t that a hellhole of complications to wrap ones head around. Broadband blues (blue chromaticities closer to achromatic) are going to be more affected by this blue-absorption-edge in terms of hue-shifts than narrowband blues! Narrowband-blues are “mostly” affected in terms of intensity but less so in terms of perceived hue-shift. At this point I’d say: never use a matrix profile for conversion of camera-rgb to whatever your connection space is. Use SSF derived LUT profiles if you can, and when in doubt, have an SSF profile for each lens.
Tried to measure some lenses today using a lamp, taking readings with and without the lens, but found the results weren’t repeatable enough. Will have to approach this differently. Might try a centered beam of light and involve some ground glass.
A diffuser would definitely help. How 'bout a piece of plywood with a 1/2" hole, with the diffuser taped to the opposite side of where the lamp shines…
Other listed cameras are coming over the next few days.
Lens data is coming once I have a repeatable measurement setup.
The camera response data and format will change as I improve the processing code until I believe I have controlled all possible factors - probably once I get lens measurement working, then camera data will be set in stone soon after.
Details of the calibration routine?
@HIRAM Calibration of what? Have you read the thread?
I’m trying stuff like this currently.
Does anyone have an idea how I could upload/back up the many GB of camera raw files? I don’t think it makes sense to do this on git. Each camera is 5-20 GB depending on compression.
In my github repo, I just upload the .csv, a .icc profile, and a README that explains the source, measurement method, and the dE report for the ColorChecker reference. I don’t think posting the source raws is worth the resources.
For comparison, I have a dataset from the ACES rawtoaces repo, sony_ilce-7rm2_380_780_5.json, which I think is the same camera as the one you posted above. Here’s its plot:
Normalized to 0.0-1.0, but is very similarly shaped to yours. They also did a multi-shot monochromator series.
Definitely don’t use git for this. Git does a very bad job of managing collections of large binary files.
The cheapest online storage you’ll find is likely Backblaze B2. It’s similar to AWS S3, but cheaper.
For online backup, I would again recommend Backblaze.
If you really want to maintain an association with the raw files in your git repo, you’ll want to host the raws somewhere like B2 and link them in your repo with git-lfs.
Thanks for the suggestions everyone. I think I’ll go with Backblaze B2, as it’s the cheapest. I don’t need any kind of linking to git. It won’t be that hard to find the data.
The GH1 surprised me. At least this explains the weird reds I’ve been getting with it (disproprtionately luminous - this issue is usually hidden by per-channel posterisation).
I’ll use the GH1 to verify if my measurements are repeatable by measuring it again later.
Will also make a webpage for accessing and viewing the data soon using github pages so I won’t have to post images.
EDIT: Just added Pentax *ist DL. It is very badly filtered and transmits sooo much UV and IR.
That’s all my cameras measured now! Also, to anyone in the UK: I would love to measure your camera and add to the dataset. Let me know if interested.
Does anyone have any idea or experience with Verity SD100 Monochromator? The one I have has Parallel outputs and serial connections in the front of the unit. i have tried to sniff out the signal path through the RS232 output to try and see what will communicate with the units etc. It is a legacy device meaning it was manufactured in 1998 but it appears to be of a high quality usable item with automatic scanning through the wavelengths capabilities as well. The only information I was able to gather was that it is a Smart Scanning Monochromator using DSP/Microprocessor controlled unit. Not much more and not sure if it take an outside controller to make it work or it that it has all of the controls inside and and can transmit useful data out etc.Actually any information would be helpful at this point.
Yes I have 3 times already at 3 separate occasions. This was a month ago and recently beginning of this week. I wish it was easy and simple that they would provide some documentation on this piece of equipment etc. Not sure what is happening to the company as it might appear to be acquired possibly with another company but not 100% sure. Thanks for the advise!
George
