aCuria wrote:
Meanwhile, the other camp swings so far the other way that they basically believe cameras are decorative accessories. According to them, image quality comes entirely from the photographer’s soul, and any sufficiently enlightened artist could produce a Pulitzer-winning portrait with a potato with a pinhole in it.
Somewhere between “this camera is ruined by 0.2 stops of shadow noise” and “gear doesn’t matter, I shoot masterpieces on a boiled russet” lies reality.
"I shoot masterpieces on a boiled russet" is a bar
I use the a9III regularly at work, and, if it weren't for P2P, I'd never notice the amount of NR that goes into those files. This is what people said they wanted; a fast camera competitive with the latest Z6 and R6. To me, from these tests and the reviews I've seen, it looks like they delivered.
NR in the same generation of sensors never added a whole stop of DR at base ISO, afaik. No FF sensor has achieved over 12 stops of measured DR even with "baked in" NR. For FF, the highest measured DR came from the The Canon R3, at 11.91 - even with NR baked in. The A1ii and A7R5 are all within that range.
This is something else. That curve is different, and the shadow improvement/ISO setting is essentially a flat line after ISO 100.
I'm intrigued to see what things look like as it gets into the wild, but Sony seems to have dipped into some genuine witchcraft for this.
freaklikeme wrote:
I'd never notice the amount of NR that goes into those files.
I suppose RAW NR could only be significant on night-sky pictures but virtually nowhere else.
q-w-z wrote:
Dual Gain is not present on this graphs, though.
They're probably doing more than just dual gain. Most of it masked by the noise reduction but there are multiple steps visible.
No, there no steps visible on contrary to most other cameras.
Dual Gain would give a setback in noise level i.e. iso 400 noise and DR will be lower than 320 or 250 (or 640-500-400 - depends on particular camera).
q-w-z wrote:
No, there no steps visible on contrary to most other cameras.
Dual Gain would give a setback in noise level i.e. iso 400 noise and DR will be lower than 320 or 250 (or 640-500-400 - depends on particular camera).
It’s confusing because several different technologies get labeled “dual gain.”
The A7V appears to use what ARRI calls a “dual-gain architecture,” which is not the same as the more common meaning of “dual gain” the dual-native-ISO system that switches to a second base ISO.
ARRI’s dual-gain architecture is essentially a specialized form of multi-frame noise reduction / HDR but instead of capturing multiple frames one after another, it captures two frames simultaneously. One is read at low gain for maximum highlight detail, and the other at high gain for cleaner shadows. The camera then combines them to produce a single raw image with lower noise and greater dynamic range.
aCuria wrote:
It’s confusing because several different technologies get labeled “dual gain.”
The A7V appears to use what ARRI calls a “dual-gain architecture,” which is not the same as the more common meaning of “dual gain” the dual-native-ISO system that switches to a second base ISO.
ARRI’s dual-gain architecture is essentially a specialized form of multi-frame noise reduction / HDR but instead of capturing multiple frames one after another, it captures two frames simultaneously. One is read at low gain for maximum highlight detail, and the other at high gain for cleaner shadows. The camera then combines them to produce a single raw image with lower noise and greater dynamic range. ...Show more →
What you refer to sounds like the dual output gain implemented in one of my cameras, Panasonic G9II. The sensor output is processed through two parallel circuits with different gain levels — one at lower gain and one at higher gain. Then, the camera merges these two into a single 16-bit RAW. Note, however that the PDR of G9II is presented by circles, not inverted triangles on photonstophotos, unlike the data points for A7V. Another thing is that the PDR of G9II doesn't change when switching from mechanical to electronic shutter. The dual output gain works with both. There is a shutter speed limit, however. The dual output gain on G9II requires SS faster than 1s, from 0.8s. The change/difference is huge. When taking dark frames with SS 1s and 0.8s, the former is hot with noise, the latter is nearly black when the exposure is pushed by several stops in post. Thus, while it might be interesting to compare dark frames from A7V with mechanical and electronic shutter, it is also worth comparing dark frames at increasingly longer(slower) shutter speeds.
rhawidantas wrote:
Except that it is not their findings but a summary of the different links pointing to the DR increase.
I have doubts about the claim in Petapixel regarding the dual output gain in Panasonic S1 II, "This approach works exceptionally well, but only with the mechanical shutter or the electronic first-curtain shutter — not with the fully electronic shutter".
There's no ES data for S1II on photonstophotos, unlike for G9II, that has the same PDR with both mechanical and ES when using the dual output gain.
And, the ES PDR of A7V is worse then the PDR for A7IV, although it isn't clear whether the latter was obtained with mechanical or electronic shutter, or both are the same on A7IV.
ruthenium wrote:
Wow, shocking.
I am not sure I understand the drop, ...and the data point are still shown as inverted triangles, as indicating noise reduction.
Regarding NR, that's what I expected. It's one of the reasons A7V beats S1II in DR department (the other being 33 mpx is my guess ).
If there's no indication, then it means that mechanical shutter was used on ptp, I think.
Lower DR than its predecessor is understandable, since the latter does not "waste" any sensor space on bumping up readout speed (and that's without NR!) .
.