This is a continuation of a series of posts on the Sony a7RIII. You should be able to find all the posts about that camera in the Category List on the right sidebar, below the Articles widget. There’s a drop-down menu there that you can use to get to all the posts in this series.
Like its predecessors, the a7RIII has quite a few shutter modes. In the past, many of those modes have knocked the digitizing precision of the camera fro 14 to 13, or even 12, bits. That happens with the a7RIII, but the rules have changed and the camera stays in high precision modes in more cases.
You can’t get 14 bits of precision in the a7RIII without being in uncompressed raw mode. Switching to compressed raw drops the precision to 13 bits at best. That’s not the end of the world, or even much of a problem, as we’ll see in the following post.
Before I get to the read noise (RN) and engineering dynamic range (EDR) versus ISO setting curves, which I’ll do in the next post, let me show you a few histograms so that you can see what the camera is doing. They are all dark field images (of the back of the body cap) at ISO 100 and 1/1000 second shutter speed.
The first is with EFCS and single shot (SS) mode:
This is a nice-looking 14-bit histogram. It’s a little asymmetrical, but on the whole, it’s what you’d expect to see.
Now I’ll turn on the electronic shutter, which Sony calls Silent Shooting on this camera.
There is no change in the histogram. This was not the case in the a7RII. In that camera, the precision in Silent Shutter mode was 12 bits. Nice going, Sony.
Now, using EFCS again and SS, but with compression on:
The precision drops to 13 bits, and the amount of noise stays about the same.
If we turn on the Silent Shutter and go to continuous mode compressed, here’s what we get:
Now we have 12 bits of precision and more noise.
If we turn on the EFCS and stay in compressed and continuous modes, nothing changes:
What about continuous mode and uncompressed RAW?
I was wondering the same… but that scenario was not tested.
Why would a camera maker build these shortcomings into their cameras. Wouldn’t it be easy enough just to add a faster processor and backplane?
None of that stuff is free.
But the raw compression decimates bits anyway so how can you tell its due to the ADC? That raw compression needs to go (mostly for the adaptive sigma delta part). Once you know to look for the artifacts you start to see them everywhere. I suspect it was one of the engineers pet projects and its built into the processing pipeline now.
I don’t understand what you mean by “decimates”. I don’t think you are using the term in the signal processing sense, since craw does not perform decimation if you use that meaning. The tone curve part of the craw compression truncates to 13 bits, but that is usually only visible if the RN is very low. At higher levels, it can cause holes in the histogram. The delta mod piece can, and does, remove outliers.
Typed that without looking closely at the article. I meant the tone curve and decimate as in remove a portion of. You can’t really equate it to bits because precision is lost only at higher values so it all depends on your level of exposure. In this case a dark exposure should be fine but had you exposed a white wall the difference would have been greater.
Jim,
Here is a quick comparison I did of the various A7rIII shooting modes for the electronic shutter, to give a photo representation of the data you presented. For those viewing this composite, keep in mind these are pushed +7EV, so the differences may not be representative of the typical post-processing you might perform.
https://photos.smugmug.com/photos/i-CXSJrLg/0/f669a88e/O/i-CXSJrLg.jpg
New reader here.
I need to some basic education about the graphs. Do you have a pointer to an histogram-101 text? Specifically, I do get why there is a loss of 1 bit, and then 2 bits, compared to the first histogram: this is because of the 2-wide and 4-wide holes in between the histogram pikes.
But how do you tell the very first is 14 bits?
Values are around (and above) 512, so I can assume the whole range can go up to 1024 (10 bits). Should I assume (from unsaid previous knowledge) that it actually goes up to 16384 (14 bits)?
It doesn’t sound like you need it. Read on.
Actually, 1- wide, and 3-wide, but yes.
The maximum is 2^14-1.
The range is larger than that if the histogram is not zoomed in.
You should. Well, one less than that, but that’s plenty close enough.
See? You understood it all along.
🙂
Thanks
And sorry for all the off-by-one bugs 😉
JD
Was it a typo in the 3rd paragraph? I think you meant A7RIII, instead of A7RII. Am I right?
You are correct. It’s fixed now.
Thanks.
Thank for taking the time to test.
So the main difference is in using single shot or continuous shooting. EFCS or silent shooting impacts nothing in this scenario.
Would really appreciate the answer to the first question here also: Does continuous shooting impact the bit depth when using uncompressed RAW?
Many thanks
Den
No.
https://blog.kasson.com/a7riii/sony-a7riii-edr-vs-iso-setting/