Author Topic: Danne's crop_rec_4k experiments for EOS M  (Read 84229 times)

2blackbar

  • Senior
  • ****
  • Posts: 480
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3025 on: April 24, 2021, 09:54:30 PM »
aaah... 10 years with magic lantern and im still a noob  :P

far.in.out

  • Freshman
  • **
  • Posts: 68
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3026 on: April 30, 2021, 03:40:55 PM »
How do I enable ISO higher than 3200? I can turn it on in canon menu but it falls back to 3200 set via ML. In ML when I set it higher than 3200 it returns back to 3200 after a fraction of a second.
EOS M (was 600D > 50D)

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3027 on: April 30, 2021, 04:21:57 PM »
Turn off iso buttons up and down. Under customized buttons.

far.in.out

  • Freshman
  • **
  • Posts: 68
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3028 on: April 30, 2021, 06:25:10 PM »
Thanks. Can you use a shutter of 1/24 in full sensor 1080p mode? I was able to enable it for 5K frtp with full range option, but it doesn't seem to work for 1080p.
Has anyone tried to record a pinhole video? It looks quite interesting.
EOS M (was 600D > 50D)

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3029 on: April 30, 2021, 08:04:45 PM »
Fine shutter tuning?

far.in.out

  • Freshman
  • **
  • Posts: 68
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3030 on: May 01, 2021, 08:58:24 PM »
Hey. I've been recording more of pinhole videos (natural outdoors daylight) which are quite low light and I have a few questions about ISO and noise. It looks like ISO 400-1600 produce less noise than ISO 100. Can that be true?
Is there any difference in terms of noise levels between the full sensor mode (3x3) the binning mode (1x3) and the 1:1 mode?
Is there anything one can do to help such low light scenarios?
EOS M (was 600D > 50D)

siatitel

  • Just arrived
  • *
  • Posts: 1
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3031 on: May 12, 2021, 08:12:52 PM »
I have a cuestion regarding the pixel binning methods.
Some time ago i try to find the specific code in which the binning was made. The idea was to define my own binning methods. But i am not an expert or good enough to figure it all for myself.
But i came to some conclusions, as i recall.
1 - The binning was not doing in code. At least not in c code.
It maybe is made in canon script code (less possible). Or it is made in one of the auxiliary chips and the c code only pass the image to that other chip, and the chip returns the image already binning. And thats why the binnings are only 1x1 (no binning at all) 1x3 and 1x5 (and theyre combinations like 3x3 or 3x5) because they are harcoded in hardware and magiclanter only is only taking advantage of the use of that chip. So make another binning methods or matrices are not possible.

Is it right?

2 Some one can implement his own binning methods to bin the image in c code, but the processor it's not fast enough to process the images.
Its true?
Do you think the processor could process... i dont know  maybe 16 fps, 24 fps, or not even 1 fps?

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3032 on: June 26, 2021, 07:35:12 PM »
A test version is up here:
https://bitbucket.org/Dannephoto/magic-lantern_jip_hop_git/downloads/crop_rec_4k_mlv_snd_raw_only_2021Jun26.EOSM202.zip

Might work or might not work better with HDMI out in usage. I added this change according to theBilalFakhouri:
https://www.magiclantern.fm/forum/index.php?topic=25784.msg235805#msg235805

Primarily test hdmi out and see if it gives less corrupted frames in general. If better then go on and test general recordings and check if files are fine as well without hdmi out. Check for focus pixels and other anomalies in the image.

far.in.out

  • Freshman
  • **
  • Posts: 68
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3033 on: June 26, 2021, 11:50:47 PM »
Hey Danne. Could you please make a copy of 5k frtp preset but with resolution adjusted to 960x1620 and with correct preview?
I can set this resolution but the preview is still showing 1280x2160.
EOS M (was 600D > 50D)

ZEEK

  • Senior
  • ****
  • Posts: 320
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3034 on: June 27, 2021, 03:44:23 AM »
hey Danne, NEW 2021 Jun26 Build tested on the EOS M & Feelworld F6 Plus Monitor. Good news!

Focus pixels are very muted in the middle of the display monitor hard to see which is good.

After recording a sequence of 2-minute clips at 1080 3x3, NO dropped frames at all, ZERO.

5K FRTP Mode works, made sure to record overblown highlights and overexpose for the test. NO dropped frames for 3 minutes and a few other clips

With the other 1:1 modes e.g. 2.5K, 2.8K etc half the screen is pink/purple towards the right side and there is no real-time correct preview, they're all Anamorphic stretch. When recorded, the image is fixed properly on the MLV File and there is no half pink/purple frame on the image. There are dropped frames.

So for 1080/5KFRTP users, it's all good  :)
EOS M - EOS M2 - 5D2

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3035 on: June 27, 2021, 04:22:36 AM »
Thanks for the feedback ZEEK. 1080p and 5k frtp, are they better working than with than the former build?
Focus pixels moved from former build?
1:1 modes are behaving different now?

Sorry for the follow up questions but I have no way of testing atm.

ZEEK

  • Senior
  • ****
  • Posts: 320
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3036 on: June 27, 2021, 05:47:49 AM »
Yes 1080 & 5K FRTP are working perfectly so far better than the previous build. I don't get dropped frames anymore, not even one in 1080. Previous builds were much more frequent with dropped frames in 1080 and sometimes in 5K FRTP.

I never had an issue with the focus pixels showing up on display but they're not as visible, if at all.

Yeah with the 1:1 modes show produce a half-purple screen, but it's no big deal as they have very frequent dropped frames.
EOS M - EOS M2 - 5D2

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3037 on: June 27, 2021, 07:45:08 AM »
That's a good indication for 1080p and 5k frtp.
I will have to take a look at the 1:1 modes in a week or so. Thanks

far.in.out

  • Freshman
  • **
  • Posts: 68
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3038 on: June 27, 2021, 10:14:32 AM »
Hey guys. Sorry for distracting you from working on the more important and useful features.
But can you help me make this preset - 5k frtp copy at 960x1620 with correct preview?
I remember that there was some guide for making presets?
I want this preset because I got a 7.5mm lens and I can work with a higher crop factor, and a smaller frame size works better on my slower SD card (70% bitrate reduction, I think) and I get more recording time. Maybe this would be useful for someone else too?
EOS M (was 600D > 50D)

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3039 on: June 27, 2021, 10:51:26 AM »
Hard work and nothing I can provide atm.
What we got now are hard coded presets handmade by theBilalfakhouri. Building what you want includes some heavy tinkering among those registers.
If you really want it done I suggest you talk to bilal and dig in those regs.

Or else run the 16:9 preset and tape your cam ;)

far.in.out

  • Freshman
  • **
  • Posts: 68
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3040 on: June 27, 2021, 11:10:09 AM »
So it's that much complicated...
OK. Is there an easy way to add black bars/overlays instead of taping the cam?)
EOS M (was 600D > 50D)

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3041 on: June 27, 2021, 11:28:59 AM »
Not if you haven't found out yourself by now.

ZEEK

  • Senior
  • ****
  • Posts: 320
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3042 on: June 27, 2021, 02:42:16 PM »
Tested the previous Feb build. Got dropped frames in 1080 mode.

2.5K/2.8k look good on the monitor as the EOS M Liveview, no purple frame issues on the monitor.

They look nice and clean, although "INFO1" button for framing when connected to the monitor doesn't work. ML Framing doesn't work in general with monitors.
EOS M - EOS M2 - 5D2

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3043 on: June 27, 2021, 03:05:15 PM »
To make it clear over here in my vacaybrain. 1:1 is working with older builds but not in the test build  8)?

ZEEK

  • Senior
  • ****
  • Posts: 320
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3044 on: June 28, 2021, 01:19:26 AM »
Yes that's right.
Older builds produce corrupt frames in all modes.
1:1 Real-time Framing is correct, no half purple screen on the monitor.

Newer build, 1080/5K FRTP work with no dropped frames. 1:1 modes are Anamorphic-stretched like the old Anamorphic mode and half of the monitor shows pink on the right side although this is not recorded in the file.

Also Framing is disabled, even when pressing info1 when connected to the monitor to switch between the framings.
EOS M - EOS M2 - 5D2

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3045 on: June 28, 2021, 07:42:23 AM »
I think it's safe to assume framing preview isn't an option with hdmi out by now ;).

I will test some when home again around different streams and see if 1:1 is getting better.

ZEEK

  • Senior
  • ****
  • Posts: 320
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3046 on: June 28, 2021, 09:22:34 AM »
Alright, so the previous monitor I used was the Feelworld F6 PLUS which caused the issue mentioned, this time I tested the Feelworld F5. Everything is fine now. Using the F6 again, oddly it's now fine on the Monitor, no graphic issues.

All modes display correctly as on the EOS M Screen. Just dropped frames for the high resolutions as expected. 1080 & 5K FRTP are perfect for monitoring. Thanks a lot! :D
EOS M - EOS M2 - 5D2

far.in.out

  • Freshman
  • **
  • Posts: 68
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3047 on: June 30, 2021, 11:42:53 AM »
That's really great news! I was just thinking about getting a monitor too.
What is the highest fps EOS-M can do for every mode (full sensor/1x3/1:1)?

BTW, can I make a donation? How would I do that, and to whom?)
EOS M (was 600D > 50D)

ctfire

  • Freshman
  • **
  • Posts: 53
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3048 on: August 05, 2021, 02:45:43 PM »
try last build in 1080p, and only when sound off it's free of pink frame when sound is on footage get couple of pink frames, not many 3-4 for 4 minutes of footage.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #3049 on: August 05, 2021, 06:13:59 PM »
So the test build still causes corrupt frames?