a7S high ISO claim - - -over hyped or not ?

Started Jul 3, 2014 | Discussions thread
Rishi Sanyal
Rishi Sanyal dpreview Admin • Posts: 809
Re: a7S high ISO claim - - -over hyped or not ?

andrewD2 wrote:

I'm not sure what you're trying to say, but if one camera has more base ISO DR than the other for one scene, it'll also have higher base ISO DR for any scene, no matter what the DR of that scene.

Apologies in advance if I'm misunderstanding what you're trying to say.


A dynamic range measurement has to have a cut off at which you say 'this isn't good enough quality shadow detail to be counted as useful DR'. In other words you can say, this camera has 13stop DR of a high quality, or it could be measured as 15 stops DR if the quality threshold is lower.

I understand that. DxO uses a cutoff of SNR=1. You may or may not agree with that cutoff, but it does account for read noise difference between cameras. Higher SNR cutoffs for DR measurements lead to DR measurements being more & more affected by shot noise - providing less and less discrimination between similar-size sensors with different read noise performances. Ultimately telling you more about sensor size, not sensor performance.

In other words, there's some good logic behind using low SNR cutoffs. Incidentally, our SNR cutoff is ~3 for our JPEG tone curve tests. We could probably debate cutoffs till the cows come home, which ultimately may be part of the reason DxO settled on 1.

DXO marks DR results are hopeless, a 5DIII still registers some info at a similar DR to a 6D but the banding of the 5DIII makes the last few stops of DR of very poor quality. Some of the Sony sensors show good detail but the limitation is colour accuracy.

I wouldn't call them hopeless, as they correlate quite well to usability of shadow detail in RAW files. For example, according to their data, base ISO DR goes as follows: a7R > a7S >> 5DIII. Which correlates exactly with my real-world experience. You do bring up a good point about banding, which make the 5DIII shadows even less useful. That's a tough problem to tackle, though.

On a 15 stop scene it maybe they (A7R and A7S) both pull up acceptably noisy detail from the deepest shadows and the higher megapixels resolves more detail. However if you pulled up detail from a more extreme scene (or even the same scene shot underexposed) it maybe (I'd have to test it) the lower megapixel camera is still showing some detail whereas the higher megapixel camera gives you nothing useful.

No, because while it's theoretically possible that the lower megapixel camera might have more DR b/c of higher FWC (if pixel-level read noise remains the same), the whole point of the RAW DR test is to see if this is actually the case. In this case, read noise swamps the lower signals on the A7S, so it cannot match the performance of the A7R (at base ISO).

The particular read noise swamping low signals on the A7S at base ISO is typically referred to as 'downstream' read noise - b/c it's introduced downstream of the ISO amplification step. The reason the A7S DR is higher at higher ISOs is b/c amplification of the signal off the sensor decreases the impact of this downstream read noise. That is, upstream read noise becomes the dominant factor limiting SNR of lower signals (ignoring shot noise, b/c this is similar across similar-sized sensors). This is noise introduced upstream of the ISO amplification step - typically the read noise of the imaging sensor itself. This upstream, sensor-level, noise is rather low on the A7S, but it's pretty low - at the pixel-level - on the A7R and 5DIII as well. The aggregate upstream read noise, however, is likely lower on the A7S due to it having fewer pixels. Furthermore, due to some other optimizations, it seems the upstream read noise may be even lower on the A7S than the A7R and/or 5DIII at the highest ISOs as well. Both of these factors (fewer pixel count, lower pixel-level read noise) would contribute to the A7S pulling ahead at higher ISOs.

But at base ISO - or, put another way, for very low signals *not* further amplified before digitization - the A7R wins. And that means it has more base ISO DR than the A7R for any scene.



-- hide signature --

“It is a capital mistake to theorize before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts.” -- Sherlock Holmes

 Rishi Sanyal's gear list:Rishi Sanyal's gear list
Sony RX100 IV Nikon D810 Canon MP-E 65mm f/2.5 1-5x Macro Nikon 85mm F1.8G Sigma 35mm F1.4 DG HSM Art +3 more
Post (hide subjects) Posted by
(unknown member)
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow