Sorry. Incorrect.
The 1DXIII uses either or of the processors for AF, and not both.
it uses one for EVF shooting and the Digic X for live view.
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
The 1DXIII uses either or of the processors for AF, and not both.
it uses one for EVF shooting and the Digic X for live view.
How'd you get that?
And that's REALLY close. Probably same or similar CFA.
Cool!
Super thanks
A few things comes to mind...
1. cost
2. IP... who knows how many variants Sony patented.
3. fab, and processes
4. Time and engineering man power limitation
5. Features for future releases.
6. goes with point 5... maybe reserved for something like the R1 if there is one?
I think you are incorrect.
This quote is from the white paper released by Canon: " And, the dedicated DIGIC 8 processor works in tandem with the central DIGIC X processor for the powerful new Head Detection AF."
Source: https://media.the-digital-picture.c...EOS-1D-X-Mark-III-Still-Image-White-Paper.pdf
In life view mode it didn’t use the digic 8 at all from what reviewers state.
And for EVF it probably uses the digix ex for head detection processing. But the AF is still driven by ten digic 8.
I remember thinking exactly like you till I came across what I am saying above. It stuck with me. But I can’t be bothered searching for it. At least I can’t think of a simply search phase to easily find it.
But think about it. Who cares about head detach when you have eye defect in live view? Unless this is specifically for EVF shooting. And there you have limited AF coverage, and information for subject recognition etc etc.
lightnchade
Leading Member
- Messages
- 584
- Reaction score
- 613
Many thanks Iliah. Looks promising.
Anyway, the 1DXmark III uses two processors to perform some autofocus functionality, which was my initial point. However, we don´t know the consequences of using two vs one processor for autofocus yet.
Iliah Borg
Forum Pro
Download code from https://github.com/LibRaw/LibRaw
In src/metadata/canon.cpp, immediately after the line:
case 2024: // 1D X Mark III, ColorDataSubVer 32
add the following line:
case 3656: // EOS R6, ColorDataSubVer 33;
Compile LibRaw
Run:
bin/raw-identify -w <filename>
Funny Valentine
Senior Member
- Messages
- 1,394
- Solutions
- 1
- Reaction score
- 1,948
because it doesn't give that much advantage. Maybe a 1/6 of a stop at best.
Last edited:
I know DPR says it's not BSI, however, are they sure about that? Nobody has said either way...
Now granted, it's not a big difference for a large surface area like the R5/R6 (FF).
This is also not a Sony fabbed sensor which the Nikon, Panasonic and Sony's are, and hence why they have BSI and the Canon does not. I gather this is not BSI as such.
However comma, Canon themselves has done some fun things here with DPAF II in relation to pixel sensitivity...
https://www.dpreview.com/forums/post/64124660
.
Likewise, if you take the JPEG engine out of the equation, the 1DX III (R6) sensor is hanging with the A7 III, (presumably) without "star eater" sensor level NR...
https://www.dpreview.com/reviews/im...t=1&x=-0.6925639426076107&y=1.058071145764048
https://www.dpreview.com/reviews/im...1&x=-0.7480349344978165&y=-0.5440107836113044
BSI is an acronym and buzz-word/term around here that is Sony tech, for lack of better words.
However, DPAF II is a (new) term by Canon, but it represents multiple patents to change how pixels and the microlens overlaying it behave which Canon is touting as class leading DR and about a stop of ISO noise reduction... And based off the DXOMark and Photons to Photons tests, is true for DR at least and is a notable ISO improvement.
I'll see your BSI, and raise you DPAF II
Last edited:
lightnchade
Leading Member
- Messages
- 584
- Reaction score
- 613
BSI's usefulness is inversely proportionate to the pixel size.
The R6 has relative large pixels so BSI is not worth the added complication and cost.
However, a more recent secondary benefit is that BSI enables stacked sensors which is useful for some features related to readout speed. But it seems Canon has figured out its own way of dealing with that anyway.
Again... the dual processor for AF is only for OVF... that is it. Which is my point. It was always difficult to fully assess DSLR vs MILC AF performance considering the wildly different concept. Different AF sensor, not all the light going to said sensor, and the lower resolution of the sensor etc etc.
here: https://www.techradar.com/reviews/canon-1dx-mark-iii
where they state “That’s not the only imaging engine available on board. The viewfinder autofocus system gets its own Digic 8 chip, so full AF/AE adjustment is also available at the top speed of 16fps.”
that's not exactly accurate.
the DIGIC 8 is used for DLSR AF and AE. not for the 1DX Mark III's liveview AF/AE. This is confirmed specifically in the 1DX Mark III's whitepaper.
So in other words, for Liveview on the 1DX Mark III and for the R6, AE/AF are the same.
Last edited:
BSI would be a help - not really for sensor efficiency which would be minor but for corner casting and vignetting. I would suspect that if they used BSI we'd see better lens test scores for vignetting from the RF lenses.
Canon has decided not to do BSI - the fabrication is more difficult than doing just a front side sensor.
However, if Canon decides to go stacked sensors, they must do BSI anyways - so I'm sure they are working on it.
juanmaasecas
Senior Member
- Messages
- 1,635
- Solutions
- 3
- Reaction score
- 1,268
How do you know that that cfa works for you? How did you test it? Just curious.
juanmaasecas
Senior Member
- Messages
- 1,635
- Solutions
- 3
- Reaction score
- 1,268
I had a Canon mount samyang 14 2.8 that had a horrible color cast in the corners with my old a7r2, and it didn’t show it when I tested it with the eos R. The a7r2 is BSI, the eos R is not.
Have to pull up the post but a contributor around here did a RAW comparison and it’s a 99.99%+ match. Literally.
Lulich or something along those lines was the poster.
sobrien
Senior Member
- Messages
- 1,763
- Reaction score
- 1,391
lightnchade
Leading Member
- Messages
- 584
- Reaction score
- 613
Analysis of the colour swatches from the studio scene.
You can sometimes pick out some aspects directly from the scene image on DPR, but that may include some unknown processing artefacts so it's best to download the file and play around with it.
But as an example, look at the cyan here and note the noise on the 1DX III image compared to the others. This (is one thing that) suggests a CFA with different characteristics.
However, what has me most excited about the 1DX III sensor is how nicely the red is being captured.
When you look the raw default rendering (adobe) on DPR it has an interesting but dull copper tint, but when rendered using variety of styles, it quite effortlessly adopts a wide range of rich shades.
It's the reds on Canon sensors that historically give me the most trouble and the 1DX III raw looks seems like a lovely file to work with, so far. That's not something you can always say about the reds in the raw from some Canon sensors (I've use quite a number) (and it's not always red that's problematic).
I'm obviously not taking about every shot, most Canon cameras produce excellent images most of the time, it's when you have an image that requires some extra work, that when what's happening under the hood starts to surface, sometimes in bizarre ways.
It's really quite a minor point, until it's not, and then it's not.
Last edited:
+1Analysis of the colour swatches from the studio scene.
You can sometimes pick out some aspects directly from the scene image on DPR, but that may include some unknown processing artefacts so it's best to download the file and play around with it.
But as an example, look at the cyan here and note the noise on the 1DX III image compared to the others. This (is one thing that) suggests a CFA with different characteristics.
However, what has me most excited about the 1DX III sensor is how nicely the red is being captured.
When you look the raw default rendering (adobe) on DPR it has an interesting but dull copper tint, but when rendered using variety of styles, it quite effortlessly adopts a wide range of rich shades.
It's the reds on Canon sensors that historically give me the most trouble and the 1DX III raw looks seems like a lovely file to work with, so far. That's not something you can always say about the reds in the raw from some Canon sensors (I've use quite a number) (and it's not always red that's problematic).
I'm obviously not taking about every shot, most Canon cameras produce excellent images most of the time, it's when you have an image that requires some extra work, that when what's happening under the hood starts to surface, sometimes in bizarre ways.
It's really quite a minor point, until it's not, and then it's not.
Question though...
Which Canons did you have trouble with?
For me, it's anything post-DIGIC5. I just "gave up" as you either stay in the past, with it's color but it's old tech, or, take Canon's newer colors which still good, but not as good; but you get the technological advancements.
The 1DX III sensor appears to strike a good balance though on the reds from the samples I've seen.
I thought the M6 II did too, just its AWB had hiccups (fairly frequent, which I have not seen with the 1DX III samples), which are easily fixed in post though. I gather the DIGIC X cures that problem.
Last edited:
Similar threads
Keyboard shortcuts
- f
- Forum