Author Topic: Danne's crop_rec_4k experiments for EOS M  (Read 154529 times)

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7240
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2925 on: December 12, 2020, 05:31:52 PM »
Yes, usually possible to gain a few more fps but not continuous. I choose to maintain a somewhat continuos preset.

2blackbar

  • Senior
  • ****
  • Posts: 492
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2926 on: December 12, 2020, 05:36:29 PM »
Maybe Zeek would get more time in 3K with 24fps and his 512 GB card that was fastest so far... If its better ill consider buying it.
Ive got 128gb sandisk pro with 64MB/s write speed in bench 192hz , his 512 has 68 in 240.
But his is weaker in 192hz with 62 MB/s , i get same 64 result in 192 and 240 but i cant recrd anythin in 240 anyway, it stops.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7240
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2927 on: December 12, 2020, 05:50:14 PM »
Maybe, maybe not. There's a line where increased fps will affect overall sd card speed no matter how fast the overclocking is.

2blackbar

  • Senior
  • ****
  • Posts: 492
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2928 on: December 12, 2020, 06:00:52 PM »
Zeek had higher data rate in 240hz , from this post:
https://www.magiclantern.fm/forum/index.php?topic=12862.msg231732#msg231732      Mazybe he would get 10 seconds in 24fps 3K mode .
If he needs info on how to do it + just go to Movie panel in ML, touch presets to expand it (dont press set) and there you have advanced mode, find register 6014 and change it to the left so its - 300, you have 23 somethin in 3K mode, you can go to -310 for 24fps.
I think his card would get better and longer record time.

AaLIEz

  • Just arrived
  • *
  • Posts: 1
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2929 on: December 13, 2020, 05:22:06 PM »
Does anyone still maintain and update the ML of Canon EOS M2?

ZEEK

  • Senior
  • ****
  • Posts: 329
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2930 on: December 13, 2020, 11:47:49 PM »
Maybe Zeek would get more time in 3K with 24fps and his 512 GB card that was fastest so far...
Well after testing a whole heap, my 2 cents is that the 240Hz patch is not reliable enough. In 2.8k 24p 240Hz I got a max of 8 minutes & now testing 3K 24p override, I got a max of 14 seconds with 240Hz patch. However after a few recordings it will go down to about 3/4 seconds with the same patch.
A break is needed and then it will pick up the pace again, but then the same pattern will occur. So then again....
I still think though 512GB is an amazing SD if you travel a lot and plan on shooting heaps of 2.5/2.8K Raw footage. 😎👍
EOS M

Walter Schulz

  • Contributor
  • Hero Member
  • *****
  • Posts: 8257
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2931 on: December 14, 2020, 12:28:35 AM »
Overclocking doesn't work well with all card types. I have some cards performing superb without overclocking and failing if put to the edge. You have to run some in-cam benchmarks to see how stable a card reacts. If a card stalls/stumbles often and unpredictable while benchmark runs (and it will only happen in write cycle): Nope, won't do with this frequency.
My good old 128 GB 95 MB/s Extreme Pro is outperformed by some mere Extreme in 240 MHz.

baberidge

  • New to the forum
  • *
  • Posts: 7
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2932 on: December 14, 2020, 08:55:39 AM »
Hello! Thank you to all involved in this wonderful project. I have noticed something odd when attempting 5k FRTP using Zeek's settings for continuous video, with the Nov 20 build. The recorded video seems shifted to the right, compared to the framing I see on the monitor on the back of the camera. I thought I would mention this here. I have not yet compared this with another build yet but I will attempt with the Oct 24 build.

Thanks again to you all!

baberidge

  • New to the forum
  • *
  • Posts: 7
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2933 on: December 14, 2020, 09:00:17 AM »
Ah, I meant the Oct 23 build, but I can't see it available for download. I will use 1080p MCM Rewire instead, this works fine!

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7240
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2934 on: December 14, 2020, 02:10:02 PM »
How much shifted would you say? Could you post a few visual examples? Unable to find time to do tests atm.

Skinny

  • Member
  • ***
  • Posts: 143
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2935 on: December 14, 2020, 03:54:20 PM »
Hi guys! I have one theory about overclocking.... do you know why exactly card fails? I mean, if someone with oscilloscope can watch the signal on SD pins, maybe clock edges are not strong enough, or data-edges...
sometimes, in digital electronics, waveform could be improved if you add pull-up or pull-down resistors, they could be installed in the microSD to SD adapter for example.

I can't check this myself because I only have 5D2 and 550D, but just had this idea right now, who knows, maybe it is something useful, or maybe not. Maybe card fails because it cannot complete write cycle in time, physical flash technology limitation, and there is nothing we can do. ALTHOUGH.... maybe under/over-voltage could affect write times... I think it's interesting.

baberidge

  • New to the forum
  • *
  • Posts: 7
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2936 on: December 14, 2020, 06:35:44 PM »
How much shifted would you say? Could you post a few visual examples? Unable to find time to do tests atm.

Thank you for the reply Danne! I hope this helps, I've had to delete most of the footage as I needed the space, but in this shot the subjects nose was in the centre on the camera screen, but it's to the left in the recorded video with extra space on the right that was not visible when framing the shot.
Fullsize: https://letschat.org.uk/M11-1503_frame_1170.png

And in this one, the top of the hot water bottle was in the centre, with both sides visible. There's about 40mm missing on the left and extra space on the right (the object is about 450mm end-to-end, shot with a 35mm lens): 

Fullsize: https://letschat.org.uk/M11-1519_frame_1.png

Walter Schulz

  • Contributor
  • Hero Member
  • *****
  • Posts: 8257
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2937 on: December 14, 2020, 06:40:34 PM »
<obsolete>
Thanks, baberidge!

baberidge

  • New to the forum
  • *
  • Posts: 7
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2938 on: December 14, 2020, 06:51:24 PM »
Please edit. Use thumbnail preview linked to full scale pic.

Oops, sorry. Is the above acceptable?

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7240
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2939 on: December 14, 2020, 07:05:48 PM »
Tested shortly and not very hard but seems ok to me. Are you selecting ratio from the Movie tab and not from under raw video menu? Could help to know exactly what settings were used here.

If you could perform a more revealing test it would probably show the issue more clearly. For example. Do a photo of your liveview screen, and then record a short mlv and post your examples here.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7240
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2940 on: December 14, 2020, 07:13:56 PM »
HereĀ“s an example with ratio 2.35:1

Cell phone screenshot of the screen


Mlv output as shown by Mlv app


Looks close enough to me but feel free to show me otherwise.

baberidge

  • New to the forum
  • *
  • Posts: 7
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2941 on: December 14, 2020, 07:19:44 PM »
Tested shortly and not very hard but seems ok to me. Are you selecting ratio from the Movie tab and not from under raw video menu? Could help to know exactly hat settings were used here.

If you could perform a more revealing test it would probably show the issue more clearly. For example. Do a photo of your liveview screen, and then record a short mlv and post your examples here.

Aha! I believe I selected the ratio from the raw video menu, so operator error. I'll perform the test for you a little later though, just to rule anything out. I've got some editing to do at the mo as I re-shot everything in MCM 1080p Rewire which looks fantastic :D I chose 5k anamorphic last time as I'd had some problems with aliasing when the subjects were wearing shirts with rough stitching but there were no such artefacts today. Thanks again - for the response and the project - we have literally no budget at all and wouldn't be able to do what we're doing with such gorgeous footage without Magic Lantern.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7240
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2942 on: December 14, 2020, 07:24:03 PM »
Happy to see the results later. Please share if possible.
Aliasing less with HD 1080p(mcm rewired)? Must be a lucky coincidence. 1x3(anamorphic modes) will outperform HD 1080p per always I would say ;). Especially on aliasing side.

baberidge

  • New to the forum
  • *
  • Posts: 7
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2943 on: December 14, 2020, 07:32:51 PM »
I will do! I don't think I explained myself correctly before, MCM Rewire had aliasing on the rough-stitched shirts shot with a 35mm f3.5 SMC Takumar stopped-down to F8 on a sunny day, which is why I chose 5k anamorphic for the interior shoot yesterday. However when I re-shot today (we were in a hurry) there was no aliasing with MCM Rewire, I'm guessing because of the fine stitching on the blouses they were wearing and the wide-open old zebra Pancolar I used providing a touch of softness to mitigate it.

masc

  • Contributor
  • Hero Member
  • *****
  • Posts: 1927
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2944 on: December 14, 2020, 07:33:38 PM »
I also tested 5K frtp here, and also in 16:9. I get (in the MLV) what I see (on the liveview).
5D3.113 | EOSM.202

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7240
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2945 on: December 14, 2020, 07:38:56 PM »
Great masc.
Yes hd 1080p is not bad. Most shots will work just fine. Zeek uses this setting a lot. To me 3x3 binning fake sharpens footage too much for my liking but I have been comparing ml footage way too long now  :P 8).

baberidge

  • New to the forum
  • *
  • Posts: 7
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2946 on: December 14, 2020, 07:46:15 PM »
Thank you masc, I really do think this is my error but I'll certainly do the tests and replicate what I did wrong!

The 5k 1x3 does have a delicious creaminess to it but I'm more than happy with today's results!

amitkattal

  • New to the forum
  • *
  • Posts: 17
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2947 on: December 31, 2020, 11:46:11 AM »
Does anyone know how to remove focus pixel maps if I use dng files workflow in davinci resolve ?

masc

  • Contributor
  • Hero Member
  • *****
  • Posts: 1927
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2948 on: December 31, 2020, 11:52:32 AM »
Does anyone know how to remove focus pixel maps if I use dng files workflow in davinci resolve ?
Only in the step when transcoding to DNG, so in MLVApp or MLVFS. Resolve can't remove that.
5D3.113 | EOSM.202

dron47

  • Just arrived
  • *
  • Posts: 1
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2949 on: January 01, 2021, 11:10:15 AM »
Hello everyone. In 2020 I decided to start videoshoting with eos M. To set magic lantern for it, I need a 2.0.2 firmware, but I can't find it anywhere on the internet. It just doesn't exist. Can someone give me some link to download it or send 2.0.2 archive, please??? I just can't use my canon with latest firmware(