Interesting if Sony is doing a dual gain thing similar to what nobody has mentioned - the Mavic 4 Pro. The main camera scored higher DR measurements than even the a7sIII using its auto ISO function to create what DJI calls dual fusion ISO. Same concept others are talking about, it's mashing two ISO's in real time to create an image with less noise and more DR. The Mini 5 does the same thing, and it really works.
After shooting with the Mavic 4 for a couple months, I hope more do this, the results are incredible with FF beating DR from a M43 sensor. Don't care about baked in NR, IQ is better in every way with no loss in detail, who cares about what's happening under the hood.
Yay! All is right in the universe again (well except that baked-in NR in A9III, but I think we can let it pass since it's "the first" FF GS) .
I suppose, since this is not a wide-spread technique (yet), Bill was caught off guard. I'd also expect FFTs to look somewhat different between NR and DGO, but that was only a hunch .
A74me wrote:
Paid off is more like it. Ive never regarded his testing as legit.
Your scepticism is unwarranted. Do you mind sharing what makes you think this way? Asking because Bill's charts have always offered the most conservative DR scores, notably lower than manufacturers' claims or DxO tests. Basically the opposite of what a "paid off" lab would do.
And the recent revision is a warranted one, because combining outputs of a dual-gain system, while obsiously reduces noise and boosts DR, is not a form of algorithmic noise reduction which comes at the expense of detail. Therefore the A7V sensor does not deserve the "triangle penalty" in the charts. IMO Sony did a phenomenal job here and deserves a round of applause.
The question that I don't know the answer, but probably some of you do: Could this DGO-technique considered to be 'a first baby step' into direction of computational imagery known from the smartphones where they stack up many images to sum/average them and get better dynamics, less noise etc.?
Kalainen wrote:
The question that I don't know the answer, but probably some of you do: Could this DGO-technique considered to be 'a first baby step' into direction of computational imagery known from the smartphones where they stack up many images to sum/average them and get better dynamics, less noise etc.?
In my understanding, it's more or less the same/similar idea, but the implementation is different. DGO is a low-level technique I'd say (you get the end result in a ~single exposure time), while what phones do is merging full exposures (i.e. you need to wait longer, so that multiple exposures are done).
old-gregg wrote:
Your scepticism is unwarranted. Do you mind sharing what makes you think this way? Asking because Bill's charts have always offered the most conservative DR scores, notably lower than manufacturers' claims or DxO tests. Basically the opposite of what a "paid off" lab would do.
And the recent revision is a warranted one, because combining outputs of a dual-gain system, while obsiously reduces noise and boosts DR, is not a form of algorithmic noise reduction which comes at the expense of detail. Therefore the A7V sensor does not deserve the "triangle penalty" in the charts. IMO Sony did a phenomenal job here and deserves a round of applause....Show more →
look at tonys test on noise compared to the a7iv its not even on par but way behind letalone infront. sorry but im a real world shooter not an a spec sheet shooter. i will wait till CineD tests DR the proper way " using images zoomed in "
A74me wrote:
Paid off is more like it. Ive never regarded his testing as legit.
---------------------------------------------
old-gregg wrote:
Your scepticism is unwarranted. Do you mind sharing what makes you think this way? Asking because Bill's charts have always offered the most conservative DR scores, notably lower than manufacturers' claims or DxO tests. Basically the opposite of what a "paid off" lab would do.
And the recent revision is a warranted one, because combining outputs of a dual-gain system, while obsiously reduces noise and boosts DR, is not a form of algorithmic noise reduction which comes at the expense of detail. Therefore the A7V sensor does not deserve the "triangle penalty" in the charts. IMO Sony did a phenomenal job here and deserves a round of applause....Show more →
@A74me@'s comment was not "scepticism", it was an accusation of unethical conduct. @A74me@ does not provide any evidence to support his accusation. The internet is full of such commentary, and, unfortunately, the FM forums are not immune.
Though in the spirit of this group I feel compelled to wonder if you could do better by not boiling the russet as I imagine the coherence of the pinhole would be easier to maintain in the raw state...
Shooting RAW is nice, but let’s not ignore the advantages of shooting boiled:
- Much lower thermal noise (it’s already cooked).
- Zero rolling shutter… because boiled potatoes don’t roll.
The boiled-russet workflow is criminally underrated.
snapsy wrote:
I know Bill well and that's both untrue and unfair. These are just cameras - let's keep it civil.
we have already seen test images with the a7v compared to the a7iv and the real world images never lie. 16 stops of DR
based on noise is a joke. i wonder if my radios signal will be stronger and have a higher DR with squelch turned up 🤣
ive been saying for year that the testing system was floored, it was just a matter of time.
Based on your S1II testing, this technique doesn't seem to necessarily be limited to mechanical shutter/EFC, correct?
Regardless, since this works on "partially stacked" architecture, it makes me wonder if the next step for Sony is applying it to their stacked sensors.
With an unwarranted assumption that the same 2.3x readout penalty would apply, a future A1iii could have this as an option. A stacked sensor with a 3.91ms readout could be 8.99ms with a "DR Boost" enabled. That would be right in line with the OM-1, which still has minimal rolling shutter in practice.
A74me wrote:
we have already seen test images with the a7v compared to the a7iv and the real world images never lie. 16 stops of DR
based on noise is a joke. i wonder if my radios signal will be stronger and have a higher DR with squelch turned up 🤣
ive been saying for year that the testing system was floored, it was just a matter of time.
You have your own imagination and will say whatever you want to say. Where did you see that Bill said there is 16 stops of DR? Also what's the floor value you are referring to? If you knew what you are talking about, you would at least be precise. You can't even formulate your thought in a mathematically and logically correct expression.
tctmp wrote:
You have your own imagination and will say whatever you want to say. Where did you see that Bill said there is 16 stops of DR? Also what's the floor value you are referring to? If you knew what you are talking about, you would at least be precise. You can't even formulate your thought in a mathematically and logically correct expression.
If you design a car and say thats it will reach top speed of 200klm ,dont you take it for a drive on the test tack to confirm the maths was right 🤔 are you guys actually inventors or just sit there with your log tables 🙄 if camera "A" takes an image cleaner than camera "B" dont you think your maths correspond 🤔but the problem is that the charts are the opposite , Im glad your not my engineer, even though my engineer has no idea as well, picked him up 3 times this week 😎 for wrong loadings.
A74me wrote:
If you design a car and say thats it will reach top speed of 200klm ,dont you take it for a drive on the test tack to confirm the maths was right 🤔 are you guys actually inventors or just sit there with your log tables 🙄 if camera "A" takes an image cleaner than camera "B" dont you think your maths correspond 🤔but the problem is that the charts are the opposite , Im glad your not my engineer, even though my engineer has no idea as well, picked him up 3 times this week 😎 for wrong loadings.
That's why you have a reputation here of completely incoherent.
You complained Bill was paid off, and you mentioned two points, (1) 16 bit DR, (2) artificial floor. Where did he say 16 bit DR, and what's the floor value?
Don't put out other BS that's irrelevant to those two points that don't make sense to anybody other than you. You either put out direct proof of your two points, or admit you are a wacko again.
tctmp wrote:
That's why you have a reputation here of completely incoherent.
You complained Bill was paid off, and you mentioned two points, (1) 16 bit DR, (2) artificial floor. Where did he say 16 bit DR, and what's the floor value?
Don't put out other BS that's irrelevant to those two points that don't make sense to anybody other than you. You either put out direct proof of your two points, or admit you are a wacko again.
Tony N must be totally wacko then, havnt you watched his comparrision utube 🙄watch the part where HD 120 flogs 4k 120 on the new sensor, there also goes the theory that more pixels are better and more pixels produce less noise 🤔 its as daft as your yacht can sail at 22deg apparent. and my yacht can sale at 26 deg apperent, but my sails at 40 across the ground and yours sails at 45, my yacht is going to flog yours in a race 🤣
These threads, where we collectively try to understand new technology—its characteristics and its limitations—are perhaps the best part of FM. They help us understand the products being sold to us in an unbiased manner, cutting through the marketing. They bring many users together in a way that genuinely increases shared understanding.
However, as everywhere on the internet, these threads can also get messed up. They become tangled with irrelevant comments, responses to those comments, and the arguments that follow.
So what to do about it? The best way to keep the thread and the discussion clean is simply this: when we see irrelevant comments from uncompetent users, we just ignore them completely and don’t respond at all. That way they wither away and don’t end up derailing the discussion. We can’t avoid such comments entirely, but the next best thing is to just ignore them. With this strategy, we can protect the original discussion and keep it on track—which ultimately benefits everyone the most.
Kalainen wrote:
These threads, where we collectively try to understand new technology—its characteristics and its limitations—are perhaps the best part of FM. They help us understand the products being sold to us in an unbiased manner, cutting through the marketing. They bring many users together in a way that genuinely increases shared understanding.
However, as everywhere on the internet, these threads can also get messed up. They become tangled with irrelevant comments, responses to those comments, and the arguments that follow.
So what to do about it? The best way to keep the thread and the discussion clean is simply this: when we see irrelevant comments from uncompetent users, we just ignore them completely and don’t respond at all. That way they wither away and don’t end up derailing the discussion. We can’t avoid such comments entirely, but the next best thing is to just ignore them. With this strategy, we can protect the original discussion and keep it on track—which ultimately benefits everyone the most....Show more →
the only thing you need to look at is the images or videos posted, you dont need to understand anything, its a photography forum. the problem is everyone thinks its a maths exam, its not use your eyes and commonsence.
.
.
.