Matt Grum wrote:
Because people still aren't getting this subtle point, there's a huge difference between a linear 12-bit encoding and a non-linear 11-bit encoding (used in Sony's RAW compression) which is actually preferable. The non-linear encoding uses a greater bit depth where it's needed most i.e. the shadows, and fewer bits where it makes no difference, i.e the highlights. A linear 12-bit encoding just throws away 2 bits everywhere.
Matt, your posts always make me chuckle. It seems like you're a full-time defender of Sony's stupid lossy RAW compression scheme. Your defenses of it are so spirited that it almost seems as if you'd invented the algorithm yourself
Aug 04, 2015 at 06:15 PM
Mark Metternich Offline • • • • • • Upload & Sell: On
I will be using a Metabones IV with Canon glass on a Sony A7R II. Can you elaborate on what you mean by a centered adapter? Are you talking about how flush it is against the camera body? Do you think the Sony A7R II will have the same issue using adapters?
Lastly, how does this decentering show up in the real world? What tests, specifically, can we do to test for decentering?
I don't want to hijack this thread so maybe I should start a new post. I, and I am sure others, would appreciate your feedback.
Thanks!
-Brian
I meant how parallel the mounting flanges are in relation to each other. It's not decentered in a way a lens would be decentered as there is no glass inside the adapter.
jhinkey gave me a tip on a good micrometer which will make my life easier when checking this stuff. Alternatively you can square your camera to a resolution chart and check the extreme edges. (Must use a well-centered lens)
Non-parallelism of the lens mount to the sensor can be seen with uneven sharpness across the frame.
My D800 showed this with my 135/2 APO - this lens was sharp enough wide open and the DOF small enough that my right side of the frame is slightly unsharp when the center is critically focused. If I use the right side to achieve focus then the left side was now out by a lot. Thought it was the lens.
Put the same 135/2 APO on my A7R with the (luckily?) very parallel Fotasy F to E adapter and the image is critically sharp across the entire width of the frame.
The D800 will go in for an AF adjustment AND a mounting flange adjustment sometime soon.
goto_dengo wrote:
Matt, your posts always make me chuckle. It seems like you're a full-time defender of Sony's stupid lossy RAW compression scheme. Your defenses of it are so spirited that it almost seems as if you'd invented the algorithm yourself
It's a clever RAW compression as there are no visible artifacts in real world picture taking. The only stupidity is from people getting obsessive about something that has zero impact on their pictures.
jhinkey wrote:
Non-parallelism of the lens mount to the sensor can be seen with uneven sharpness across the frame.
My D800 showed this with my 135/2 APO - this lens was sharp enough wide open and the DOF small enough that my right side of the frame is slightly unsharp when the center is critically focused. If I use the right side to achieve focus then the left side was now out by a lot. Thought it was the lens.
Put the same 135/2 APO on my A7R with the (luckily?) very parallel Fotasy F to E adapter and the image is critically sharp across the entire width of the frame.
The D800 will go in for an AF adjustment AND a mounting flange adjustment sometime soon....Show more →
The D800 had some very serious off center focus point issues. It was downplayed a lot but was a big issue. I sent mine in for repair. It came back okay but not perfect.
Non parallelism is difficult to measure because setting up a camera so the sensor is perfectly parallel with a flat target is very difficult (a mirror won't help you as as you can't guarantee the sensor is perfectly centered on the lenses optical axis). I typically do it therefore on infinity scenes where this is unimportant. The issue is greatest with wider lenses as they have the smallest depth of focus ( not to be confused with depth of field) so any manufacturing error in the lens or lens flange and sensor with regards to being non parallel will easily show. Depth of focus increases as the focal length increases so telephoto lenses can tolerate a non perfect parallel assebmly of the sensor and lens flange a bit better. As sensor density goes up - all this stuff gets more important.
gdanmitchell wrote:
Not necessarily, depending on the nature of software involved. I have no idea — which is why another test is in order — what might be going on here, but it isn't hard to conceive of software that discounts the significance of luminosity data differences that likely existed between the letters and the clouds but does not discount the significance of a similar difference that is part of a linear object between brighter and darker points.
It's not hard to conceive of, but it would be very very difficult to implement. You're talking about some sort of image segmentation / feature detection algorithm, these tasks are hard (for machines) without noisy data. When you have a significant amount of noise the difficulty factor goes up exponentially.
gdanmitchell wrote:
Yes, that is a supposition, but it is based on some knowledge of how certain kinds of data compression are accomplished, and they can occasionally produce odd and unexpected results. (Those who know how the mp3 audio compression scheme works may get what I'm saying here.)
Certain compression algorithms work in the frequency domain, and whilst it's possible to filter out low frequency (large scale features) whilst retaining high frequency noise, edges still count as high frequency content. Detecting and removing edges is again a hard problem in image processing.
To have solved some of the more difficult problems in computer vision, by accident (whilst doing what exactly?) is beyond believable I'm afraid.
Occam's Razor: letters and clouds are coincidentally the same colour so they don't show up.
shirozina wrote:
The issue is greatest with wider lenses as they have the smallest depth of focus ( not to be confused with depth of field) so any manufacturing error in the lens or lens flange and sensor with regards to being non parallel will easily show. Depth of focus increases as the focal length increases so telephoto lenses can tolerate a non perfect parallel assebmly of the sensor and lens flange a bit better. As sensor density goes up - all this stuff gets more important.
Actually I believe depth-of-focus is independent of focal length when you're focused to infinity. The formula according to wikipedia is
The magnification m depends on the focal length, however for objects at infinity the magnification is zero taking focal length out of the equation. It often seems like wides have shallower depth of focus at infinity but I don't see how this can be the case unless the formula above is incorrect (I don't happen to have the text book it's reportedly from).
goto_dengo wrote:
Matt, your posts always make me chuckle. It seems like you're a full-time defender of Sony's stupid lossy RAW compression scheme. Your defenses of it are so spirited that it almost seems as if you'd invented the algorithm yourself
I don't think you've read enough vintage Matt Grum then... since he's actually tried to engineer a way to cover up/ameliorate the very compression scheme's artifacts that you seem to think he's defending!
goto_dengo wrote:
It seems like you're a full-time defender of Sony's stupid lossy RAW compression scheme.
Nope, just a full-time defender of facts, objectivity & balance (the latter being in great demand in general online) against people who point to each and every artifact (or otherwise) in an image as evidence of compression losses. I also defended Canon and the 5Ds against the "lenses can't resolve more than 30 megapixels" crowd with equal enthusiasm, so it's not a brand thing.
goto_dengo wrote:
it almost seems as if you'd invented the algorithm yourself
If I had I would have done a better job of it. For example using non-linear delta values, and increasing the bit depth of deltas when a large brightness variation was detected. That would be sufficient to completely banish 99.9% of the artifacts (though sadly not the discussions about the fact that it's still technically lossy).
justruss wrote:
vintage
Yeah I guess I should actually get round to finishing that project sometime!
Matt Grum wrote:
Because people still aren't getting this subtle point, there's a huge difference between a linear 12-bit encoding and a non-linear 11-bit encoding (used in Sony's RAW compression) which is actually preferable. The non-linear encoding uses a greater bit depth where it's needed most i.e. the shadows, and fewer bits where it makes no difference, i.e the highlights. A linear 12-bit encoding just throws away 2 bits everywhere.
Sony pre-newbie: My impression is that a very flat, low contrast, low detail scene would have a smaller file size than a very busy, high contrast/DR scene.
I'm curious how large the file is when the lens cap is on, and 1/4000s + f22 + ISO 50. Seems that could be very highly compressed.
ldallan wrote:
Sony pre-newbie: My impression is that a very flat, low contrast, low detail scene would have a smaller file size than a very busy, high contrast/DR scene.
I'm curious how large the file is when the lens cap is on, and 1/4000s + f22 + ISO 50. Seems that could be very highly compressed.
Or not?
The compression ratio is fixed (by design) to exactly 1.75:1 so you get essentially the same filesizes regardless of image content. The exact filesizes do vary slightly, I can't remember whether that's due to the embedded JPEG thumbnail or an additional lossless compression step that is employed.
tommieh wrote:
Yes, it is entirely possible (to have Magic Lantern-like customization) since Sony has its appstore and one user found a way to create custom apps (No -- not custom WIFI-apps, custom native apps). Via such a custom app one could load an entire interface. It would be quite easy since one also can load the app via a Custom Button and the "Application list", no need to have some special card with the UI etc.
What's missing is "only" people that understand Android development and that are willing to do the work (yes -- Sony firmware uses Android).
So there could possibly be, in theory, something like an "EFSC App" (electronic first curtain shutter) for the original a7r to mostly or completely eliminate "shutter shock"?
Perhaps that app (or similar) could somehow over-ride the apparent oddity of BULB using 12-bits?
I did take a quick look at NEX-hack.info, but it appeared to be rather immature.
shirozina wrote:
It's a clever RAW compression as there are no visible artifacts in real world picture taking. The only stupidity is from people getting obsessive about something that has zero impact on their pictures.
By "real world" I assume you mean a world where selective blindness is prevalent?
molson wrote:
By "real world" I assume you mean a world where selective blindness is prevalent?
I can see perfectly clearly and can see no artifacts - can you provide some examples in your own images?
So there could possibly be, in theory, something like an "EFSC App" (electronic first curtain shutter) for the original a7r to mostly or completely eliminate "shutter shock"?
Perhaps that app (or similar) could somehow over-ride the apparent oddity of BULB using 12-bits?
I did take a quick look at NEX-hack.info, but it appeared to be rather immature.
EFCS is a hardware function (?) so programming one for the A7r would be quite hard.
Why did you check NEX-hack.info? The reference i gave was to Github, nex-hack.info is something else where development seem to have halted.
