p.4 #1 · Sony silently applying gain at certain apertures
Jonas B wrote:
The light source in this case was a LED panel made for film scanning. So, pretty bright and evenly spread rays of light, but quite intense for f/1.2. Also, not totally flicker free as can be seen in the image series.
With a filter I could have darkened the source and used a slower shutter speed and then the histogram curves would have been nicer. Well, next time... ;-)
This test is not hard at all to repeat with your own camera provided you have a fast lens with manually driven aperture and electronic contacts. Perhaps somebody with a more recent camera and a CV 40/1.2?
Thank you ruthenium for checking everything, digging into the raws... and I'll just continue thinking the increased gain takes place in the camera....Show more →
I have the A1, however my fastest lens is the 35mm F1.4 GM - not superfast. This might be worth looking into anyways, at F1.4.
After looking at your two raw files, I cannot say I have any substantially better understanding of the observations (unfortunately).
1) While the difference in lightness is apparent - it is relatively small. I cannot say that seeing this difference is immediately alarming, like seeing some real problem.
2) I cannot confidently attribute this difference to a vignetting correction. It is hard to say whether the change in lightness is or isn't uniform. To illustrate what I mean, I added two full-sized jpegs at 90% quality obtained from the two raw files from Jonas (no corrections have been applied in post). While one of the two jpegs has less dark corners, the result doesn't look like real good-quality and meaningful vignetting correction.
3) I cannot help thinking that the camera might be using the aperture together with the shutter speed to get an idea of the exposure, to decide the gain. This is pure speculation on my side, but I am not sure if this can be a problem when no information about the aperture (f-number) and the focal length is available (when the electronic communication is lost between the lens and the body). In other words, I am not sure if this is a correct comparison (with a connected and disconnected lens) that should be interpreted as indicating a silent gain or a silent vignetting correction, and not something else.
ILCE-1Voigtlander NOKTON 50mm F1.2 Aspherical lens50mmf/1.21/20000s100 ISO0.0 EV
p.4 #2 · Sony silently applying gain at certain apertures
ruthenium wrote:
[...]
2) I cannot confidently attribute this difference to a vignetting correction. It is hard to say whether the change in lightness is or isn't uniform. To illustrate what I mean, I added two full-sized jpegs at 90% quality obtained from the two raw files from Jonas (no corrections have been applied in post). While one of the two jpegs has less dark corners, the result doesn't look like real good-quality and meaningful vignetting correction. [...]
I don't think it is related to the vignetting correction. That's probably just an optical illusion. Here are the steps I went through:
Open both images in Photoshop, brighten the darken one (image 56) until histogram windows show the same RGB value for the whole image as is measured when looking at image 51.
Place image 56 as a new layer on image 51. Set the layer mode to difference.
Flatten the image and then invert it.
Now we have some very weak differences only and it's about traces from the banding only. There is no sign of difference towards the corners.
p.4 #3 · Sony silently applying gain at certain apertures
ruthenium wrote:
I don't know whether some correction (whether it is gain, or shading compensation, or something else) is applied in-camera on the raw data and thus is "baked-in", or whether there is some "soft" control written in EXIF, or mabe even both.
Questions of this kind can be non-trivial. Some information can possibly be gained by experimentation, however there is a limit to what one can do without the insider knowledge.
Thank you for sharing the files - I shall have a look at these later today.
The "baked-in vignetting correction" is applied in camera to the RAW data and not reversible. As can be seen in @snapsy's thread, too, I think. So this really should be disabled if you shoot RAW. It may be that the camera for some reason forces this to be enabled in some cases, which would certainly explain this thread.
p.4 #4 · Sony silently applying gain at certain apertures
Jonas B wrote:
I don't think it is related to the vignetting correction. That's probably just an optical illusion.
You could enable vignetting correction and compare the result. Vignetting correction should generally produce a higher gain towards the edges, whereas any other exposure bias should result in an even gain difference across the field. And don't forget to disable it again after your test:-D
p.4 #5 · Sony silently applying gain at certain apertures
Daran wrote:
You could enable vignetting correction and compare the result. Vignetting correction should generally produce a higher gain towards the edges, whereas any other exposure bias should result in an even gain difference across the field. And don't forget to disable it again after your test:-D
Nah... there is a limit to how interested I am in this matter.
I use the camera mainly for still images. When filming I don't use that fast lenses. So, from were I am everything looks fine.
I can see though that this in camera increased gain can be a problem in situations like those ImDaJim, the OP, described.
p.4 #6 · Sony silently applying gain at certain apertures
I took four shots with my A1 & 35mm F1.4 GM at F1.4, while shooting with RAW + JPG
Two shots with the lens properly connected, with Vignetting correction set to Off (first shot) then to Auto (second shot).
Two shots with the lens twisted by a touch to disconnect the electronic contact with the body, with Vignetting correction set to Off (third shot) then to Auto (fourht shot).
I converted the raw images to jpg without corrections.
The resulting jpgs are compiled as seen below. On the left are the converted raw images, on the right are the out-of-camera jpgs. The labels indicate whether the lens was connected or not, and whether the vignetting correction was Off or Auto.
Settings: full manual (f/1.4, ISO 100, ss 1/200, observed exposure 0.0), mechanical shutter, no DRO, focus at infinity, the lens was set to touch the display with neutral grey on the display. WB was set to daylight.
Observations:
1) When the vignetting correction is Auto, this correction is baked in the raw. When it is Off, then there is no vignetting correction neither in raw nor in the out-of-camera jpg.
2) When there is no connection between the lens and the body, setting vignetting correction to Auto has no effect. The resulting raw and jpgs are darker by -0.3 EV. To me, this seems insignificant for most practical reasons. Furthermore, I doubt that the difference can be unambiguously attributed to some "silent gain" added by the camera when it is connected with the lens. An alternative way of thinking is suspecting that when the camera has no information about the focal length and aperture, it cannot properly determine exposure. There might be a different explanation.
I attached the raw histograms: the top is when the lens was connected, and the bottom one is when the lens was disconnected.
p.4 #7 · Sony silently applying gain at certain apertures
ruthenium wrote:
I took four shots with my A1 & 35mm F1.4 GM at F1.4, while shooting with RAW + JPG
Two shots with the lens properly connected, with Vignetting correction set to Off (first shot) then to Auto (second shot).
[...]
So basically the same result as I got (with an A1 as well). There is a slight difference with a lens at f/1.4 depending on electrically connected to the the camera or twisted a bit.
(Earlier findings show that at f/1.2 and f/1.1 the differences are bigger and bigger again.)
I think you showed that the vignetting setting and action by the camera doesn't matter, it's just another thing going on (or preferably not going on).
