Menu

Show posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Show posts Menu

Messages - rbrune

#26
Reverse Engineering / Re: JTAG / UART & more
April 04, 2016, 09:50:10 AM
The buspirate is a fine little device.

Here is me dumping some flash memory with it: https://twitter.com/_deeperblue/status/466329008746266624

As a1ex said the flash will likely just be 1:1 the content of the ROM0/1.BIN dump files. But if you're able to read/write the flash with the chip still on the camera board (like I did in the photo - but sometimes that doesn't work due to the board layout and how power is distributed) that would open up a great way to reanimate bricked cameras. Same is true if you get JTAG working. Also the buspirate together with flashrom should autodetect the flash chip/type - if that doesn't happen there's probably a wiring issue and/or reading/writing the chip in place doesn't work due to the board layout and it's voltage distribution.
#27
Yes, the latest version allows more fine grained control over the resolution.

And for 100MB/s - maybe someone with a 5d3 can chip in but I assume something around 90MB/s is more realistic as it leaves some bandwidth and ram to cope with hickups/CF slowdowns during writing.
#28
Feature Requests / Re: mv1080 on EOSM
February 08, 2016, 03:28:56 PM
Having this implemented would be really sweet and I'm motivated to look into it.

We have to figure out a way to set the EOSM to mv1080 when starting recording in the mlv module. A starting point would be to take a look at the 650D (is this the most similar firmware/hardware?) and see what it does when switching to 1080p mode in the canon menu and see if a similar function could be called in the EOSM firmware.

And I think in a couple of places inside the mlv module it is assumed that mv1080 mode on EOSM is actually only mv720 mode - this needs fixing if we ever manage to actually switch modes correctly.
#29
Raw Video / Re: 7d 2k continuous 2.67 aspect ratio
November 06, 2015, 01:59:05 PM
None of those resolutions are 2.67 aspect ratio? Are you sure you clicked the apply aspect ratio button in my mlraw calculator?

Also you should measure your real life performance of your camera+memory card and put that value minus a little bit of buffer into the calculator to see what your setup can achieve.
#30
Raw Video / Re: 2.39 Ratio Missing Resolutions?
November 03, 2015, 11:01:43 AM
It's due to a patch/enhancement I made to decrease the amount of preset resolutions and allow for better control of the resolution. You should be able to use left/right while having the resolution selected to fine tune it in small fine grained steps.
#31
The correct scaling factor is 5/3 = 1.666[...] as it uses a binning rectangle of 3 by 5 pixels to create one output pixel.
#32
Raw Video / Re: raw canon 50d max resolution ?
August 28, 2015, 04:14:51 PM
Use this calculator to see what the 50D is capable of: http://rbrune.github.io/mlraw/
#33
Sadly the Resolve 12 Beta crashes on startup. I guess it's because of my AMD card - even though I had no issues with 11.x, maybe I have to try different drivers.
#34
Take a look at the calculator here to understand what is going on internally:

http://www.magiclantern.fm/forum/index.php?topic=14909.0
#35
General Help Q&A / Re: 5dmk2 Benchmark confusion
May 09, 2015, 09:15:42 PM
Yes it is to be expected.
The card interface of the 5D2 is limited to around 70 to 75MB/s.
#36
Quote from: Grzegorz on May 07, 2015, 11:06:04 AM
Canon 50D today, with Lexar x1066 64GB, in crop mode managed over six minutes with no frames skipped at 1728 x 1080, aspect 16:10, I stopped the recording myself. :)

Would someone pls. direct me to a thread explaining framing in crop mode? What I see is obviously not what I get...


Regarding the framing, open my calculator again and look at the rectangle that shows you the portion used of the sensor. Of course you have to comapre it to the APS-C sized grey frame to see how much/little of the full sensor area you're using. Also the calculator gives you the crop factor with which you can multiply your lens focal length with to get the effective focal length.
Also there are some settings regarding framing indicators in the ML mlv settings, play around with those.

Now 1728x1080@24fps for 6mins is about 75MB/s. Are you sure those are your settings and really no frame was dropped? I thought the theoretical limit of the 50D interface is about 70MB/s and you exceeded it by quite a bit.
#37
Quote from: Kharak on May 06, 2015, 01:13:10 PM
it says 100+20 from SD.. But I don't think I have ever gotten close to 120MB/s write speed.

120MB/s is a very theoretical limit. Do not expect to get anywhere above 100MB/s.
Either reduce fps or resolution to get longer/unlimited recording time.

You can use this to estimate bandwidth use: http://rbrune.github.io/mlraw/
(make sure you do set your measured write speed after selecting the MKIII to see what your limit for long recording times is)
#38
Raw Video / Re: Cropping issue 650D
April 22, 2015, 08:13:39 PM
Check out this calculator, it should help you understand what is going on:
http://www.magiclantern.fm/forum/index.php?topic=14909.0
#39
Quote from: vertigopix on April 20, 2015, 04:23:53 PM
Thank you !

a) Yes, max 21 MB/s with a fast card exFAT formatted   :(
b) No, max selectable crop is 2496x1080

Awesome, thank you.

@b) 2512 is probably the actual max - but mlv_rec only allows multiples of 32 (like my calculator), so everything is correct now.
#40
Quote from: vertigopix on April 20, 2015, 12:00:21 PM
Nice tool, thank you !

Maybe can you add 60D ?  :)

Done (it is basically a 600D)
.
But since I do not have a 60D, can you confirm to me that
a) the max SD card write speed is about 21MB/s
b) the max selectable resolution in crop mode is 2512x1080
#41
Raw Video / Re: 30 frames VS 24 Frames.
April 17, 2015, 08:51:39 AM
Quote from: reddeercity on April 17, 2015, 06:14:49 AM
I never really tested it out for any Production Work just to beach mark my Lexer 1066X.
If you decide to use it make sure you do Full Card Tests . I personally use feb16th nightly build
as it's the Last build with 1872*xxxx  , after that the code was limited to 1856 for 5D2.
It's the only one I trust in Full Film/Video daily Production. (Time Is Money)

It's because 1856 is a multiple of 32 and 1872 is not. mlv_rec limits the resolutions to multiples of 32 because it allows for faster memory copy - and by that also better/stable write performance.
#42
Quote from: the associate on April 14, 2015, 05:38:53 PM
Hi, any M owners out there with updated raw video capability info? Am using the new U3 SD cards from SanDisk and Transcend, and it seems I've been able to pull fairly decent resolutions. My highest continuous record for 2:39:1 at 23.97 fps is currently 1728x434 (using MLV RAW w/sound). Still nowhere near CF, and still limited by the M's 40MB write speed, but at least it seems to be an improvement. Just wondering what resolutions other M owners are pulling these days. Also wondering if most of the M info on Roman's most excellent raw capability chart still applies. Thoughts?

Take a look at this calculator of mine:
http://www.magiclantern.fm/forum/index.php?topic=14909.0

1728 is actually the limit of what the EOS-M can record, it is already using the whole width of the APS-C sensor with 3x5 sampling. The calculator will allow you to see what the practical and theoretical limits of the EOS-M and other cameras are.
#43
Quote from: dubzeebass on April 08, 2015, 02:31:13 PM
Can u add ability to calculate run time per card size? Great job!

Great idea - and done.
Added a snippet on how many minutes a 32GB card will be good for.
#44
Take a look at this calculator I made:
http://rbrune.github.io/mlraw/

Since 600D is very similar to the 550D, click on the 550D pick your frame rate and aspect ratio. You will see that the camera is also listed with a maximum write speed of 21MB/s. Regardless of the SD card you buy the camera won't be able to save faster than 21MB/s.
#45
Since the topics of maximum SD/CF card writing speeds, achievable maximum resolutions and how the different recording modes affect the crop factor come up over and over again I put together a little javascript calculator that should help understanding the capabilities of the respective cameras better - and might be useful for planning shoots.

How to use:
Select your camera model, select the desired recording mode (zoom, 1080p, 720p) which determines the pixel subsampling mode, pick an aspect ratio and frame rate. Ideally also type in the max. writing speed of your SD/CF card measured by magic lantern (otherwise it uses the limit of the SD/CF interface which might not be achievable with your card).


http://rbrune.github.io/mlraw/
(updated April 20th, 2015)

If you miss a feature or something doesn't seem right please let me know.



Disclaimer:

  • The pixel width given by the calculator might not be selectable in magic lantern on your camera since it currently doesn't offer finer grained selection (the calculator allows for all multiples of 32 while magic lantern only has a subset of those selectable).
  • I tried to include as many quirks and limits for each listed camera (max crop recording sizes, no 1080p mode on EOS-M, fps limitations of 500D, ...). Please report any inconsistencies and additional quirks!
#46
Be aware that when you shoot raw with the 600D the max resolution you will get with 16:9 is 960x540 with 24p (that is 20.8MB/s which is at the limit of the SD card interface).
And on top of that the effective crop factor in 3x crop mode will be ~8.75x - with out the 3x crop mode you will get an effective crop of ~2.92x. (that is due to the fact that you only use a subset of the sensor width due to the low resolution selected).
#47
You can edit in Resolve. Exporting to h264 is only for the end result of grading+editing. Hence the cheapest and easiest workflow is to just load the cdng sequences in Resolve, grade and edit them in there and export as h264 for delivery.
#48
To clear up the confusion about 720p mode and vertical pixels in 2.35:1 ratio: The computation of the squeeze factor wasn't always correct and a patch I made submitted recently fixes that.
The correct vertical squeeze factor is 5/3 = ~1.666x before it was reported as 1.61x
So in current magic lantern one should get a resolution of 1920/(2.35*1.66) = 490.
#49
@ a1ex & g3gg0: have you tried using buildroot ( http://buildroot.uclibc.org/ ) to build a FLAT arm init+userspace?
#50
EOS-M also seems to boot it.

Is there already a git/hg for kernel/initrd/userspace set up?