Author Topic: Danne's crop_rec_4k experiments for EOS M  (Read 85373 times)

Tullen

  • Freshman
  • **
  • Posts: 73
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2450 on: July 04, 2020, 05:17:55 PM »
Im happy I didnt have time to do my guide yet ;). Still planing on doing it though. But I guess till will move some goalposts.

Edit: Or maybe not? Danne: "The card speed depends on fps. Even if the card theoretically seems to work for this it seems with 24 fps it maxes out around 57Mb/s. This is not the case with full resolution preset which runs with a reduced framerate."

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2451 on: July 04, 2020, 05:54:25 PM »
Exactly. That´s why benchmarks are so fast when in play mode. Higher fps reduces the effect.

Tullen

  • Freshman
  • **
  • Posts: 73
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2452 on: July 04, 2020, 06:03:25 PM »
Exactly. That´s why benchmarks are so fast when in play mode. Higher fps reduces the effect.

Do you think it will translate at all for 24 and more fps? To 18fps 5k ana flv?

Kgncom

  • New to the forum
  • *
  • Posts: 18
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2453 on: July 05, 2020, 12:28:12 AM »
so, I understand benchmarks are hardly a 1:1 analogue for in the field performance, but I’m confused as to how 24fps recording would bring the write speed back down SO much
70D | EOSM

Tullen

  • Freshman
  • **
  • Posts: 73
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2454 on: July 05, 2020, 12:32:51 AM »
I guess the most straight forward is just to test and see if you get any better real world results recording.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2455 on: July 05, 2020, 10:10:52 AM »
New build:
https://www.magiclantern.fm/forum/index.php?topic=9741.msg208959#msg208959

- 1736x2930 16fps mode(16:9)

Former hidden "silent film" mode in 2.39:1 18fps now becomes richer 16:9 format. To reach this feature select 5k anamorphic flv and select set 25fps from movie tab menu. Push menu button twice and it should go into 1736x2930 16fps. In 10bit continuous with last 192Mhz patch.

Kgncom

  • New to the forum
  • *
  • Posts: 18
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2456 on: July 06, 2020, 07:45:38 AM »
I’ve never fiddled around with the anamorphic modes before but I’m wondering what exactly the difference between the rewire and not rewire is.
70D | EOSM

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2457 on: July 06, 2020, 08:45:21 AM »
Early preset. I erased one of them just now and renamed a few presets to reduce confusion. Redownload and see if you still have the same questions after this version.

guerchi

  • New to the forum
  • *
  • Posts: 19
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2458 on: July 06, 2020, 07:04:55 PM »
Sandisk Extreme Pro 128GB 95MB/s

Low level format in camera (Menu button -> First tool icon -> Format card -> tick in low level format and press OK)

A few tests runs of the benchmark as well as taking a picture, so card was not totally empty when I run the below benchmarks.


Modules tab -> Bench.mo ON -> Restart -> Debug tab -> sd overclock -> 160MHz -> Restart -> Debug tab ->  Benchmarks -> Card Benchmarks -> Quick R/W benchmark (1 min) -> press play

160MHz
Play-160mhz" border="0

Debug tab -> sd overclock -> 192MHz -> Restart -> Debug tab ->  Benchmarks -> Card Benchmarks -> Quick R/W benchmark (1 min) -> press play

192MHz
Play-192mhz" border="0

Edit: 240MHz gave around 20 MB/s and Global Draw Off made no difference.
I can't achieve those marks that you achieve, with the same Sandisk Extreme Pro 128GB 95MB/s
These are my numbers after low level formatting:
sd overclock 192MHz:

sd overclock 160MHz:

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2459 on: July 06, 2020, 07:40:37 PM »
@guerchi. Could you test this version? It goes straight into 192Mhz. I tweaked it a little. Maybe works better:
https://bitbucket.org/Dannephoto/raw2mlv/downloads/crop_rec_4k_mlv_snd_raw_only_2020Jul06.EOSM202_ver_1.zip

I run these numbers:
Code: [Select]
static uint32_t uhs_regs[]     = { 0xC0400600, 0xC0400604,/*C0400608, C040060C*/0xC040060C, 0xC0400614, 0xC0400618, 0xC0400624, 0xC0400628, 0xC040061C, 0xC0400620 };   /* register addresses */
static uint32_t sdr_192MHz[]   = {        0x8,        0x3,                             0x4, 0x1D000301,        0x0,      0x201,      0x201,      0x100,        0x4 };

guerchi

  • New to the forum
  • *
  • Posts: 19
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2460 on: July 06, 2020, 08:39:01 PM »
@guerchi. Could you test this version? It goes straight into 192Mhz. I tweaked it a little. Maybe works better:
https://bitbucket.org/Dannephoto/raw2mlv/downloads/crop_rec_4k_mlv_snd_raw_only_2020Jul06.EOSM202_ver_1.zip

I run these numbers:
Code: [Select]
static uint32_t uhs_regs[]     = { 0xC0400600, 0xC0400604,/*C0400608, C040060C*/0xC040060C, 0xC0400614, 0xC0400618, 0xC0400624, 0xC0400628, 0xC040061C, 0xC0400620 };   /* register addresses */
static uint32_t sdr_192MHz[]   = {        0x8,        0x3,                             0x4, 0x1D000301,        0x0,      0x201,      0x201,      0x100,        0x4 };
@Danne, thanks! i just tested but  I get almost the same results:

I will try with the 170MB/s card.

guerchi

  • New to the forum
  • *
  • Posts: 19
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2461 on: July 07, 2020, 02:24:57 AM »
I tested the Sandisk Extreme Pro 128GB 170MB/s and give me these results in play mode on 192MHz:

 In real escenario I can get stable 62MB/s that means Continuous 2320x1254 14bit (1.85:1) in 2.5K Mode!
When it seems that the EOSM has reached the abyss of the possible, you do it again, thanks Danne!

Kgncom

  • New to the forum
  • *
  • Posts: 18
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2462 on: July 07, 2020, 05:02:40 AM »
So toyed around with the 16fps silent mode but i was always getting corrupted purple frames. When I turned off the force 25 fps mode though the fps dropped to 14 and I wasn't getting any purple frames.
I decided to test record time in this 14fps silent mode. My resolution was 1736x3256. I was using the card that I benchmarked on reply #4211 in 192MHz overcloc k- Sandisk256GB 95MB/s. I went through the various bitrate settings (10-14) with two exposure scenarios, one as overexposed as possible, and one only a little overexposed. I also had global draw turned off in the video submenu, and sound was disabled. Once a clip hit a minute I called it continuous and stopped recording.
Here is an imgur album of six screen caps of different examples (3 for overexposed and 3 ETTR) https://imgur.com/a/W62ZSyx
It seems my card's in the field write speeds taps out at roughly 68-70MB/s. This new build is pretty stellar. Thanks Danne!
70D | EOSM

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2463 on: July 07, 2020, 08:02:37 AM »
I uploaded a new build. Could you test for corruption in 16fps in that version?
https://www.magiclantern.fm/forum/index.php?topic=9741.msg208959#msg208959

Kgncom

  • New to the forum
  • *
  • Posts: 18
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2464 on: July 07, 2020, 10:21:27 AM »
I uploaded a new build. Could you test for corruption in 16fps in that version?
https://www.magiclantern.fm/forum/index.php?topic=9741.msg208959#msg208959
Sure
70D | EOSM

Kgncom

  • New to the forum
  • *
  • Posts: 18
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2465 on: July 07, 2020, 11:11:14 AM »
I'm not experiencing corrupted frames now, Danne.
70D | EOSM

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2466 on: July 07, 2020, 11:12:57 AM »
Cool. Should be fixed.

Stousen

  • New to the forum
  • *
  • Posts: 43
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2467 on: July 08, 2020, 02:47:02 PM »
Does anyone know if it is ML that impacts my manual photo settings on my eos m?
I cannot seem to do any auto iso shots. It allways reverts back to 100.

Also, i cant choose auto iso in ml menu either, while in video mode.
 Anyone have a clue?

Cheerio

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2468 on: July 08, 2020, 02:51:49 PM »
Turn off iso button handling from crop rec menu. It's under customized buttons.

Stousen

  • New to the forum
  • *
  • Posts: 43
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2469 on: July 08, 2020, 02:59:49 PM »
Turn off iso button handling from crop rec menu. It's under customized buttons.

Oh yes! Imagesaver Danne. Thank you. Out in the woods and dont want to spend too much battery on trying iso changes.  :) :)
Best!

Cheers

rinski

  • New to the forum
  • *
  • Posts: 44
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2470 on: July 08, 2020, 06:47:59 PM »
Hello, I wanted to comment that I miss the height in the anamorphic 5k the height of 1900 as before July 5 and even more the 2160 in 16.9 with the increase in speed, in 12 bits.
Thank you.

WeakestLinkForNow

  • New to the forum
  • *
  • Posts: 8
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2471 on: July 08, 2020, 10:09:02 PM »
So i'm just getting back to using my EOS M after borrowing it out to someone for a month. Is there a way to get 2.5k without the crop?

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2472 on: July 09, 2020, 12:01:27 PM »
Hello, I wanted to comment that I miss the height in the anamorphic 5k the height of 1900 as before July 5 and even more the 2160 in 16.9 with the increase in speed, in 12 bits.
Thank you.
Increasing height will cause corrupt frames. I tested extensively and found sweet spot. This has nothing to do with faster sd_uhs patching.

IDA_ML

  • Hero Member
  • *****
  • Posts: 963
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2473 on: July 09, 2020, 12:06:38 PM »
I have now briefly tested Danne's latest (July 8-th) experimental build for the EOS-M with the 95 MB/s card.  I am happy to tell you that:

1) At 5k anamorphic (1736x2928), 10-bit lossless and 16.688 fps recording is continuous and without corrupt frames!  I get about 15 minutes of footage with these settings on a 64 GB card, (vs. about 20 min. at 1736x976@24fps) - not bad at all keeping in mind the lack of aliasing and the tremendous resolution at the full sensor readout!  Hands down - this 5k anamorphic mode at 17 fps is revolutionary!

2) Overclocking is automatic and set to 192 MHz.  Upon start-up camera is ready for use, no need to do anything.  Really nice!

3)  These additional 2.4 fps compared to the old 14.288 fps seem to provide a noticeable improvement of the jello effect in critical scenes with fast lateral motion.

4)  0 and 180 deg. shutter (1/16-th and 1/33-th of a sec.) is easily set by the Shutter fine tuning function - very important for smooth footage and perfect for low-light shots at 1/16-th of a sec. 

5)  Camera operation is very stable and works well with AF and Magic zoom enabled. 

Genious work, Danne, congratulations!

Question:
-----------
        Is it possible to implement this exact functionality also on the 100D?  There are two major reasons why I am asking this:

A. The 100D has still several essential advantages compared to the EOS-M - the optical viefinder which is much better and convenient for taking stills photos, especially at bright light, the EF-mount allowing to use EF lenses without adapters and the much longer battery life in photo mode which makes the 100D a much better choice for stills and video shooting during travel compared to the EOS-M, especially for those who take lots of stills too.

B.  For event shooting, the combination of the EOS-M and 100D which both are very light and compact cameras is very convenient.  If one of the cameras is equiped with a portrait lens for close-up shots and the other one with a wide-angle lens, this offers great flexibility for quickly changing gear according to the scene without missing valuable moments. 

And thanks a lot for your excellent work!

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7188
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #2474 on: July 09, 2020, 12:23:41 PM »
16.688 fps? Messing with fps override again, sigh. Anyway. Lucky it works, otherwise sticking to 16fps should be fairly stable to use.
100D, I left it for anyone else to maintain.