Sensor Gain and manufacturers' hidden tweaks

[photohiker]

Well this is interesting.

How many times have you heard someone complaining that the actual exposure on a DSLR didn't quite reflect the exposure settings given when compared between cameras. Sometimes more than quite. I've certainly seen it a few times, and have found annoyance that sometimes there seems to be a hidden hand altering the results. What I didn't realise (but might have if I'd thought about it) is that a numerical F-Stop on one lens is not necessarily passing the same light to the sensor as the same F-Stop on a different lens. I also didn't realise that our manufacturers are tweaking the sensor gain to compensate, behind our backs as it were...

Thankfully with digital, chimping allows us to fine tune our settings to get the desired result without the issues of film processing, but how hard is it really to do it right the first time?

Quite hard apparently:

Luminous Landscape: Open letter to camera manufacturers

Be watching for responses to that.

Michael

[surenj]

While this is interesting, I see no practical application or relevance of this information.

What's the error of their measurement? Was there a true difference in the means (of those EV values) if so, what was the confidence interval?

[digitor]

Have you just done a stats module mate?

Cheers

[surenj]

Just watched the inception. Still dreaming I think... or am I?

[Murray Foote]

I've know for quite a while about the difference between T stops and F stops and that the claimed maximum F stop of one lens is not the same as the maximum F stop of another. What I didn't realise is that sensors have a part to play too. I'm not sure that there's any tweaking goning on by manufacturers, though. One thing I noticed is that the D3 for example performs better than a D300 so that the low light performance gain of larger sensors with lower pixel densities is even greater than mere F stops would suggest.

[malificentmuse]

I remember at a photoshoot with Cameron, that we were getting different results despite using the same setting and lens on his d300 and my d90. We were using flashes as rim lights so the difference became instantly noticeable when we chimped and compared. Turns out it was because of d-lighting and when that was turned off, we both attained similar results with the set up.

[Murray Foote]

That could be another factor. The preview on the back of the camera is sRGB, much smaller than RAW. In a RAW image you will have more exposure latitude than either the preview or the camera histograms show but it's impossible to know exactly how much. The algorithm to produce jpegs or previews on the back of the camera could conceivable vary from camera to camera or from manufacturer to manufacturer.

At least one of you was shooting Jpeg to be using D-Lighting since that attempts to give a bit more of the RAW range to a Jpeg image (a different JPeg processing algorithm). I can't help wondering how they calculate T Stops from the file on the camera. Perhaps they use the RAW file without any adjustments (standard settings but not optimal settings). Perhaps they use a RAW file of a standard image adjusted to maximise tonal range through a standardised procedure. If they don't do the latter they are understating the T-Stop compared to what you can extract from the image. Either way, if you shoot JPeg, you are effectively choosing a restricted tonal range and further limiting the T-Stop (i.e. reducing your low-light sensitivity).

Hmmm. Afterthought. I wonder if there's a case for turning on D-Lighting when shooting RAW. It's not going to affect the RAW image (apart from the initial editing appearance if you're using Capture NX) but it might give previews and histograms that more accurately show the potential of the RAW image. Something to test out in a difficult lighting situation.

[chrisk]

I've know for quite a while about the difference between T stops and F stops and that the claimed maximum F stop of one lens is not the same as the maximum F stop of another. What I didn't realise is that sensors have a part to play too. I'm not sure that there's any tweaking goning on by manufacturers, though. One thing I noticed is that the D3 for example performs better than a D300 so that the low light performance gain of larger sensors with lower pixel densities is even greater than mere F stops would suggest.

i didnt know this was goign on and to be honest i still dont know if i get it 100%. all i thought that was going on was that the manufacturers were being quite liberal in their estimation of iso values at the top end.

[Murray Foote]

Come to think of it, Ive seen several camera reviews that compare the accuracy of ISO settings and some can be out by quite a way, all throughout their range. If you're using just the one body you'd probably never notice.