Author Topic: Danne's crop_rec_4k experiments for EOS M  (Read 373065 times)

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7702
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1525 on: August 17, 2019, 03:44:05 AM »
hi! made some test videos yesterday and made footage to test black sun issue, found that it not gone, there is some gray colors, that turn to pink if you try to move highlights even few steps down.
there's video about issue to show you what i mean
august 09 build, raw mcm rewire, 14 bit lossless
Well, could you upload a sample file? Also, is this happening in for instance ACR? Did you compare other NLE;s?

kuchka94

  • New to the forum
  • *
  • Posts: 17
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1526 on: August 17, 2019, 11:10:43 AM »
Well, could you upload a sample file? Also, is this happening in for instance ACR? Did you compare other NLE;s?
sorry, can't understand meaning of ACR and NLE with my english)))
uploaded 3 mlv files, all of them have same issue, so i think in every footage with direct sun it can appear.
https://drive.google.com/open?id=1YT88ZvEZdTb7PB6Z-C5Rt4lyiEzUP2hX
https://drive.google.com/open?id=17ImK5GuA-LCBS48zpiBsbli1e-FfAg9n
https://drive.google.com/open?id=13s0fGs5W2E_bSuRolxJLYuw2iLVAqpPG

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7702
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1527 on: August 17, 2019, 11:26:54 AM »
Ok, did some fast testing. Set white level to 14000 on Mlv App. It´s now 16200. This issue also appears in acr(adobe raw engine).
So these shots are done under iso 100. Is this problem also happening in iso 200 or iso 400? Might be the same issue as described here:
http://www.magiclantern.fm/forum/index.php?topic=16040.msg191131#msg191131

By the way. is 16200 a good default value for eosm? I probably should know this but if anyone can find a more correct white level default please tell...

walter_schulz

  • Contributor
  • Senior
  • *****
  • Posts: 396
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1528 on: August 17, 2019, 11:35:54 AM »
sorry, can't understand meaning of ACR and NLE with my english)))

Adobe Camera Raw: Raw converter used by Adobe. Embedded in Lightroom, too.
Non-Linear Editing system

ZEEK

  • Senior
  • ****
  • Posts: 329
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1529 on: August 17, 2019, 12:17:52 PM »
is 16200 a good default value for eosm?

16200 for almost all cases is absolutely fine! The EOS M has only a RAW Dynamic Range of 11.2 EV, so pushing it around a lot won't help. My 6D has 12.1 EV and doesn't suffer much from these issues as the DR allows for better highlight/shadow recovery.
An ND Filter & good exposure is the key here to somewhat avoid these pink highlights and a bit of tweaking with the highlights and white levels in MLV App to which I have found 14444 or 15000 to be my go-to, although not always the best options.
Attempting to recover blown-out highlights is going to cause some issues, especially with a camera that has decent dynamic range, although good enough for most cases.
EOS M

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7702
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1530 on: August 17, 2019, 12:41:25 PM »
For 100D safety margin is 13400 and for the rest isos 15200. I will eventually put in 15200. Would like to differ iso 100 to 13400 but my iso_climb function and other anomalies will make it better for an overall 15400 white level in 14 bit:
Code: [Select]
#if defined(CONFIG_100D) || defined(CONFIG_700D) || defined(CONFIG_EOSM) /* other models? */
    /* http://www.magiclantern.fm/forum/index.php?topic=16040.msg191131#msg191131 */
    /* 100 units below measured value = about 0.01 EV */
    /* due to iso_climb routines this value has to be set to a compromise */
     default_white = 15200;
   // default_white = (lens_info.raw_iso == ISO_100) ? 13400 : 15200;
#endif

ZEEK

  • Senior
  • ****
  • Posts: 329
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1531 on: August 17, 2019, 12:48:06 PM »
Thanks Danne. Yeah, a default White Level of 15200/15400 would help. The ISO climb function is genius.
EOS M

2blackbar

  • Hero Member
  • *****
  • Posts: 514
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1532 on: August 17, 2019, 03:29:42 PM »
By the way. is 16200 a good default value for eosm? I probably should know this but if anyone can find a more correct white level default please tell...
From my experiments most stable were 12 bits with white leve because it wasnt jumping to different values from clip to clipl, its 5138 for me,  i have to dial it manually in MLVApp , in 10 bits white level jumped to different value between clips from the same shoot , it was not one number most of the time, with 14bits i did not test this extensively so i settled for 12.
Tested this with highligt reconstruct and adjusted white level until i had it reconstructed without pink or magenta shift of pure white.

kuchka94

  • New to the forum
  • *
  • Posts: 17
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1533 on: August 20, 2019, 01:36:12 AM »
made tests for different iso, 100-800, on direct sun.
in every shoot black sun problem appears with white level 16200.
white level 13500 solve everything in every shoot.
for another shoots without sun white level 16200 is good.

Fadis

  • New to the forum
  • *
  • Posts: 7
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1534 on: August 20, 2019, 05:53:47 PM »
Good day to all.
I am very glad that I have such a good camera,   such a tiny size with 22mm!
I have one question - I can not achieve long, stable video recording (especially with a bright day or room) in the 1080p\rewire\14bit to the Samsung EVO 2 128g card. Are there any  settings that affect the stability of recording to the card without sacrificing quality?

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7702
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1535 on: August 20, 2019, 06:24:50 PM »
12bit, turn on kill global draw.

rinski

  • Freshman
  • **
  • Posts: 50
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1536 on: August 20, 2019, 07:45:09 PM »
Hello, please could you guide me how to decrease the 4 or 5 lines of small red horizontal dots that come out permanently after recording 4k anamorphic 1280x1866 1x3 12 bits in clipping, 2.39 format.
Thank you very much.

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7702

Naviddabest

  • New to the forum
  • *
  • Posts: 5
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1538 on: August 21, 2019, 12:26:19 AM »
Good day to all.
I am very glad that I have such a good camera,   such a tiny size with 22mm!
I have one question - I can not achieve long, stable video recording (especially with a bright day or room) in the 1080p\rewire\14bit to the Samsung EVO 2 128g card. Are there any  settings that affect the stability of recording to the card without sacrificing quality?

i used to use the samsung evo plus 128gb the faster model there is 2 versions apparently. and at times i got upto 45MBps it was not stable at all somethings did help such as using the card warmup, using prerecord and using a differnet sd apapter as mine was a micro sd. anyways best advice just get the sandisk extreme pro. the annoying thing for me was that i bought the evo plus specifically for ML as my research showed it was "good enough" but i guess i was wrong.

yokashin

  • Member
  • ***
  • Posts: 122
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1539 on: August 21, 2019, 06:01:27 AM »
i used to use the samsung evo plus 128gb the faster model there is 2 versions apparently. and at times i got upto 45MBps it was not stable at all somethings did help such as using the card warmup, using prerecord and using a differnet sd apapter as mine was a micro sd. anyways best advice just get the sandisk extreme pro. the annoying thing for me was that i bought the evo plus specifically for ML as my research showed it was "good enough" but i guess i was wrong.

https://www.magiclantern.fm/forum/index.php?topic=12862.msg208719#msg208719
https://www.magiclantern.fm/forum/index.php?topic=12862.msg201519#msg201519
https://www.magiclantern.fm/forum/index.php?topic=12862.msg201547#msg201547
70D.112 [main cam] | M.202 | S110 [CHDK]

Fadis

  • New to the forum
  • *
  • Posts: 7
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1540 on: August 21, 2019, 08:05:31 PM »
https://www.magiclantern.fm/forum/index.php?topic=12862.msg208719#msg208719
https://www.magiclantern.fm/forum/index.php?topic=12862.msg201519#msg201519
https://www.magiclantern.fm/forum/index.php?topic=12862.msg201547#msg201547

I am using MB-MC128GA. The card is fast enough, recording in the card reader is about 83 megabytes per second, but for some reason the recording is interrupted periodically in bright light. This is especially aggravated when the memory card is full more than 70 percent.

walter_schulz

  • Contributor
  • Senior
  • *****
  • Posts: 396
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1541 on: August 21, 2019, 08:14:17 PM »
File system used: exFAT or FAT32?

Fadis

  • New to the forum
  • *
  • Posts: 7
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1542 on: August 21, 2019, 09:08:10 PM »
File system used: exFAT or FAT32?

exFAT

lightspeed

  • Freshman
  • **
  • Posts: 64
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1543 on: August 22, 2019, 11:02:02 PM »
Can someone tell me how to activate a 3k setting in mv_lite without using crop_rec. I have noticed a conflict between the two when shooting in reduced bit rates.

Would I have to do it with a Lua script as that is how I activate 2.5k now.

Hmm

  • New to the forum
  • *
  • Posts: 2
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1544 on: August 23, 2019, 07:30:42 AM »
Hello everyone, I have the same problem as Fadis, same samsung 128gb, exfat. On my PC it gives very good read/write speeds but on EOSM it's very unstable and typically gives about 25mb/s.
BUT, if I reformat this card in camera then I get stable max possible speed for EOSM. What's the reason for this to happen? Why formatting helps, but after some short time card starts underperforming? Even tho on PC speeds are stable all time.

This is my old post https://www.magiclantern.fm/forum/index.php?topic=9741.msg211051#msg211051

ngemu

  • Freshman
  • **
  • Posts: 83
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1545 on: August 23, 2019, 01:16:34 PM »
Can someone tell me how to activate a 3k setting in mv_lite without using crop_rec. I have noticed a conflict between the two when shooting in reduced bit rates.

Would I have to do it with a Lua script as that is how I activate 2.5k now.

It would be nice if more scripts were included in that menu with various working settings at different resolutions, crop modes, or aspect ratios. It would definitely make the camera easier to use, especially with my peanut brain.

DarkCAL

  • New to the forum
  • *
  • Posts: 9
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1546 on: August 26, 2019, 12:23:38 AM »
@Hmm Guess that's why we all use the Sandisk Extreme Pro SD card just because of the issues you're having right now

yokashin

  • Member
  • ***
  • Posts: 122
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1547 on: August 26, 2019, 08:10:18 AM »
Hello everyone, I have the same problem as Fadis, same samsung 128gb, exfat. On my PC it gives very good read/write speeds but on EOSM it's very unstable and typically gives about 25mb/s.
BUT, if I reformat this card in camera then I get stable max possible speed for EOSM. What's the reason for this to happen? Why formatting helps, but after some short time card starts underperforming? Even tho on PC speeds are stable all time.

This is my old post https://www.magiclantern.fm/forum/index.php?topic=9741.msg211051#msg211051


Why You Shouldn’t Use MicroSD Cards In DSLR Or Mirrorless Cameras
https://www.howtogeek.com/398225/why-you-shouldnt-use-microsd-cards-in-dslr-or-mirrorless-cameras/

"This is an interesting subject, and I started digging into the SD protocol a little bit to figure out why his is happening.

One thing I've discovered is that the SD interface is a serial interface, but with a twist. Devices can use one or four wires for data. The choice of single-wire SPI vs 4-wire transfer obviously is going to make a huge difference in transfer speed. It's possible that the SD adapter you were using didn't have all of the wires connected, or there was a defect in the card, leaving a wire disconnected and causing the camera to fall back to 1-wire SPI.

A little reading on this article has also illustrated that a high number of I/O operations per second and fast sustained throughput are not the same thing, and cards can be optimized for one or the other.

Finally, I looked up the SD protocol, just to see if there's a limitation in µSD cards that I'm not aware of. Electrically, the only difference seems to be that SD cards have two ground pins, while µSD only has one.

Finally, just for fun, I opened up one of my µSD adapters to see what's inside. It's entirely a passive adapter, with no circuits or electronics - just flat, metal traces connecting the pads on the SD plug to the pins on the µSD socket.

It's possible the bridge connecting the two VSS pins (notice it up at the top) might be causing issues with the CLK or VCC line, but since VCC is the power supply and CLK is the clock, a short on either one of those would cause the card to fail a lot more spectacularly, so I don't think that's the issue."


70D.112 [main cam] | M.202 | S110 [CHDK]

Danne

  • Developer
  • Hero Member
  • *****
  • Posts: 7702
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1548 on: August 27, 2019, 01:43:00 PM »
New build:
https://www.magiclantern.fm/forum/index.php?topic=9741.msg208959#msg208959

branch:
https://bitbucket.org/Dannephoto/magic-lantern_jip-hop/branch/crop_rec_4k_mlv_snd_raw_only

Been working hard with Jip-Hop to refine presets and all around movie handling approaching the eosm more like a movie cam. Some of the things done recently:

- reworked Movie tab menu(Jip-Hop)
- fast preset access by simply tapping display
- corruption free 4k anamorphic modes(slightly reduced height in raw.c)
- shutter coherency. Same shutter across all presets(fixed a bug introduced by myself)
- Movie Restart for raw recordings(Jip-Hop)
- autoenabling modules(refined by Jip-Hop)
- gain buttons up for higher iso. push down for lower iso, also working while recording
- aperture + ISO(up/down buttons will open or close aperture when max/min iso is reached)(Jip-Hop)
- h264 mov preset, working with x3 digital gain(SET button) and gain buttons(up/down)
- toggling between photo and movie mode working better

Upon install buttons will be reassigned as follows:
SET x3 crop and back to x1.6 mode
Up/Down gain increasing/decreasing(iso)
Tap display startoff preset list
Halfshutter press x10 zoom aid(release for normal view)

For newcomers and for people who just wants to get down to recording do following:
1 - Install above build onto your eosm and start film. Camera will be set to mv1080p 2.39:1 24 fps
Or...
2 - Set your camera into manual focus mode and tap your display to get into the fast access preset list:

Select a preset after tapping your display. Done!:


Reworked Movie tab menu(thanks Jip-Hop):


Movie Restart feature(thanks Jip-Hop)


customize buttons:


h264 mov. Select it from tap display options:


Missing 3K, 4K? Still there. Reach them by tapping "presets" from Movie tab section:


cine 2.39:1 script still there(refined by Jip-Hop)


Enjoy hard open source work!

As always greatful for a1ex sharing his architectural magic lantern code work. Of course g3gg0 and others not mentioned too...

yokashin

  • Member
  • ***
  • Posts: 122
Re: Danne's crop_rec_4k experiments for EOS M
« Reply #1549 on: August 27, 2019, 02:03:39 PM »
My highest commendation  :)
70D.112 [main cam] | M.202 | S110 [CHDK]