Author Topic: Danne's crop_rec_4k experiments for EOS M  (Read 275092 times)

roterabe

  • New to the forum
  • *
  • Posts: 4
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3300 on: September 20, 2022, 08:49:34 AM »
New build:
Missed a few pixels on the right. Fixed now!

Isn't the new build referencing the 5D Mk.3? It could still be the right one, but I'm not that into the project to know what boot check determines on what camera I'm trying to run this.

Edit: I'm assuming this is related to build options? I'm not seeing anything in the commits which would change the behaviour to be incompatible with EOS M

cedricp

  • New to the forum
  • *
  • Posts: 28
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3301 on: September 20, 2022, 09:38:11 AM »
Working here too :) I had to compile it by myself cause the binaries are for 5D3 ;)
Doing some tests right now

cannibalferox

  • New to the forum
  • *
  • Posts: 14
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3302 on: September 20, 2022, 09:41:23 AM »
@cedriccp whats the command for compilation? Is there an instruction? Or could you host it?

cedricp

  • New to the forum
  • *
  • Posts: 28

cannibalferox

  • New to the forum
  • *
  • Posts: 14
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3304 on: September 20, 2022, 09:49:51 AM »
thanks a lot!

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7503
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3305 on: September 20, 2022, 09:52:09 AM »
I replaced to eosm build.

masc

  • Contributor
  • Hero Member
  • *****
  • Posts: 2051
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3306 on: September 20, 2022, 09:53:54 AM »
@Danne! Thanks a lot. So crazy... 5.2K anamorphic on M became reality. Damn. That's great. In my first 2 tests I got no pink frame. Setting: your 1736x2178, but 12bit. MLVApp told 79MB/s and 81MB/s in my first two tests. With 10bit it should be continous also in bright scenes.
You guys are great, bilal & danne!
5D3.113 | EOSM.202

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7503
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3307 on: September 20, 2022, 10:01:17 AM »
Great.
Yes. 10bit is the way to go  8)

cedricp

  • New to the forum
  • *
  • Posts: 28
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3308 on: September 20, 2022, 10:24:41 AM »
Awesome, tested with wide lens (and speed booster), it's amazing. Using Resolve AI upscaler add even more sharpness to the image.
Great job dudes  8)
I'm wondering what we can expect more from that little beast

cannibalferox

  • New to the forum
  • *
  • Posts: 14
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3309 on: September 20, 2022, 11:04:11 AM »
 @Danne what does the preview vs real-time setting in the screenshots above help with? On first glance the output looks exactly the same if I use the info button choppy uncropped preview or is switching with the info button turning on preview. No offense just want to understand what I am missing here.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7503
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3310 on: September 20, 2022, 12:24:54 PM »
Nothing missing. Same functions activated.

cannibalferox

  • New to the forum
  • *
  • Posts: 14
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3311 on: September 20, 2022, 03:13:43 PM »
Okay I was just confused if your screenshot instructions would have something totally new I have missed :D 5,2k quality is brutal ! If it would be possible to get at least centered and lets say 80% of the real fov this would be insane

Edit: What trickery is behind 16:9 5k and 2.35:1 frtp mode? :) Or would it be possible decrease the size of the stutterfree liveview as in 2.5k preset and then display the whole fov?

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7503
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3312 on: September 20, 2022, 03:43:40 PM »
You can try center with cmos5 or cmos7 in the submenu.

cannibalferox

  • New to the forum
  • *
  • Posts: 14
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3313 on: September 20, 2022, 04:34:09 PM »
sorry Google doesnt show me a decent instruction where this submenu is or what it is called and how to switch :/ I only know the possibility to turn on canon gui and use the arrows.

Edit: Found it by accident I think? Play button on the presets if extended with advanced settings has these registers

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7503
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3314 on: September 20, 2022, 04:54:12 PM »
Way harder tasks have been solved in this place. Take your time  8).

cedricp

  • New to the forum
  • *
  • Posts: 28
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3315 on: September 20, 2022, 05:05:05 PM »
Interesting, is there any calculation doc about registers settings for cmos5 and cmos7 ?
I started with values taken from the code (0x20 & 0xB87) but it's not clear how it works.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7503
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3316 on: September 20, 2022, 05:17:16 PM »
There's some logic but easier to try small steps and simply check visually. I never got any further than doing that.

gabriielangel

  • Member
  • ***
  • Posts: 102
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3317 on: September 21, 2022, 01:05:03 AM »
Shameless bump  8)

New version out:

Works well for me. Able to record 3 minutes consistently at 10bits. Thanks!

Ernani

  • Freshman
  • **
  • Posts: 50
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3318 on: September 21, 2022, 02:03:16 AM »
Okay I was just confused if your screenshot instructions would have something totally new I have missed :D 5,2k quality is brutal ! If it would be possible to get at least centered and lets say 80% of the real fov this would be insane

You can get it centered almost perfectly by tapping on screen arrows. First turn off "tap display" on custom buttons (otherwise it wont work). Then put the preview in real time ant tap the arrows until it matches the real framing preview. You can get is perfectly centered horizontally. Vertically, you will get very close, just a bit skewed towards the top.

likwidchz

  • Just arrived
  • *
  • Posts: 1
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3319 on: September 21, 2022, 03:03:45 AM »
New user here, happened to hit refresh on a page and saw there was a new build.
Tried it out, the 5k worked at ~1:23 seconds before the buffer filled up @ 14 bits.
Will have to try how long it does constant recordings, looks like @Gabriielangel mentioned 3 minutes. Did the camera auto stop or did you stop it?

Thanks!

Btw, do people see any visual difference between 10-14 bits? or is it a wash?

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7503
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3320 on: September 21, 2022, 09:05:03 AM »
You can get it centered almost perfectly by tapping on screen arrows. First turn off "tap display" on custom buttons (otherwise it wont work). Then put the preview in real time ant tap the arrows until it matches the real framing preview. You can get is perfectly centered horizontally. Vertically, you will get very close, just a bit skewed towards the top.
Nice tip 👍

cannibalferox

  • New to the forum
  • *
  • Posts: 14
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3321 on: September 21, 2022, 09:41:49 AM »
yes thats what I also tried but after hitting left arrow key a bunch of time the camera has frozen :D but I also wanted to experiment with the arrows today.

BatchGordon

  • New to the forum
  • *
  • Posts: 14
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3322 on: September 21, 2022, 10:32:56 AM »
Being a 5.2k, this new setting has so much vertical resolution that dual-iso is no more so much penalizing in terms of loss of detail.
Danne and Bilal did a wonderful job!

julien becker

  • New to the forum
  • *
  • Posts: 3
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3323 on: September 22, 2022, 01:23:39 PM »
Hi,

Danne and Bilal what you offer to the ML community is amazing ! Thank you so much.

I've done some benchmarks with my SD cards and I'm not sure to understand what happens. I've first tested the SanDisk R170 128GB in 5.2K mode (with the settings provided by Danne).
If I launch the benchmark in video mode I get around 40 MB/s and malloc_errors (I don't know what these are). If I do the benchmark in stills mode, I get 68MB write speed and 63MB read speed.
Is that not a bit low ? I've also tested an Sony R260 MB/s W100 MB/s and those figures were even worse, like 19 MB/s.

https://drive.google.com/open?id=1Z41X7BytwAIjmjtOdI982npknXsgWPRZ&authuser=julien%40skilllab.net&usp=drive_fs

When I record on the SanDisk, it works continuously for over 6 minutes (I've stopped it). So I guess that's the most important, but I would like to understand the banchmark and these malloc_errors.

Cheers

Walter Schulz

  • Contributor
  • Hero Member
  • *****
  • Posts: 8539
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3324 on: September 22, 2022, 01:45:08 PM »
I've done some benchmarks with my SD cards and I'm not sure to understand what happens. I've first tested the SanDisk R170 128GB in 5.2K mode (with the settings provided by Danne).
If I launch the benchmark in video mode I get around 40 MB/s and malloc_errors (I don't know what these are). If I do the benchmark in stills mode, I get 68MB write speed and 63MB read speed.
Is that not a bit low ? I've also tested an Sony R260 MB/s W100 MB/s and those figures were even worse, like 19 MB/s.
a) Make sure to use Danne's latest build linked in first post of this thread.
b) Highest transfer numbers are only visible in "Play" mode. You will find a line beginning "Mode: PLAY-UNK LCD" in top section. If you get something like "Mode: PH-LV LCD" or "Mode: MVC-1080 LCD" you are in liveview mode. To enter Play mode you have to press Play button immediately after starting benchmark run.
c) If you constantly get malloc errors in benchmarks you should unload mlv_*** modules (if available) and restart camera. Or disable raw/mlv recording.
d) If a card cannot cope with selected overclocking frequency transfer will drop to 20 MByte/s (give or take) which is lowest transfer mode. Select a lower frequency and restart your camera.
e) R260 indicates an UHS-II card. EOS M has UHS-I interface and your card will work in UHS-I compatibility mode and given transfer speeds are not valid here. Without test nobody can say how good any UHS-II card will do in this mode. Furthermore nobody can say (without test) how a card will react to overclocked UHS-I mode.