Author Topic: Canon 70D  (Read 1437686 times)

andy kh

  • Senior
  • ****
  • Posts: 326
Re: Canon 70D
« Reply #3075 on: January 12, 2018, 09:25:15 AM »
Can't test as i am out of station at the moment. MlvProducer has the option to change the black level. It happen when 10bit/12bit raw was implemented in 70D

fjodor

  • Just arrived
  • *
  • Posts: 2
Re: Canon 70D
« Reply #3076 on: January 12, 2018, 10:10:00 PM »
Unfortunately 14bit-lossless is still broken in my tests. I use MLVP v3200. Perhaps the lossless compression in this build is a bit better than the previous build (see attachment), I do not know ...
Also the mlv_sound module seems broken in this Jan11-build and wont boot-up. To be honest I have never tested a 70d-build where the sound from mlv_sound was synchronous and / or continuous in a logical way with raw video. My workflow is MLVP > C-DNG > Resolve. Maybe I'm doing something wrong and someone has some good tips for usable audio.





I'm hopeful that lossless compression will come to the 70d, hopefully with usable audio, fingers crossed :-p
Thanks again for everyone's hard work and commitment to ML

David_Hugh

  • New to the forum
  • *
  • Posts: 8
Re: Canon 70D
« Reply #3077 on: January 13, 2018, 06:08:18 PM »
Just tested the January 11th build from esas. Unfortunately, I have to report that the files still come out like this:

 This was 12 bit lossless, global draw off/ highest resolution possible. 14bit looks the same. Does your 14 bit lossless come out differently @esas? The right blacklevelfor MLV files in MLV Producer is 127 by the way. Changing that is no problem with the new MLV Producer version, it works with the lossless files.

Fazit: Everything "works" but those pink "stripes" (more like bars really) arent exactly pretty ;). Is there anything else to test?

farrellts

  • New to the forum
  • *
  • Posts: 7
  • Yada yada yada bada bing bada boom!
Re: Canon 70D
« Reply #3078 on: January 17, 2018, 12:06:30 PM »
I was reviewing the list of Magic Lantern features which cannot be implemented on EOS 70D due to firmware limitations.  Okay, I don't know a thing about the CPU on this device or what language its firmware was written in, but I would imagine that somebody had to run a decompiler on the firmware at some point, while verifying what limitations it presents when porting Magic Lantern to run on it.  Furthermore, I would imagine, that if it could be demonstrated that the source code of decompiled firmware was sound enough to be compiled and link-edited to create machine code identical -- or at least functionally identical -- to the original firmware from which that source code was decompiled -- well, then I'd imagine the source code could then be tweaked so that the desired missing features of Magic Lantern would then run on EOS 70D.

Now, I know that's a lot of ifs -- and testing tweaked firmware would not only void warranty but could damage the camera.  But I imagine it could be done -- unless there is an actual hardware limitation on this camera beyond what the firmware will allow.  It might be foolish to try this ... but I cannot help wondering if somebody has?  Anybody in this forum perhaps?

As a software developer myself with pretty broad experience, I know the challenges and dangers inherent in a project like this.  It's certainly not something I would undertake alone.  I've worked with some very impressive decompilers, but have never seen one generate perfect source code which can be flawlessly compiled.  Perhaps if one wants higher-fidelity video, it might just be easier to go buy one of the increasingly-affordable 4K-capable cameras out there?  But I'd still like to pull more out of my 70D.

I'd be interested in comments on this. Hope I am not going wildly off the range of topics which this forum hosts?

Please note that it is not my intention to disparage Magic Lantern as implemented on 70D.  I truly appreciate the work involved in creating and maintaining this build!
Tom Farrell 70D.111A

Walter Schulz

  • Hero Member
  • *****
  • Posts: 5878
Re: Canon 70D
« Reply #3079 on: January 17, 2018, 12:17:47 PM »
ML does not alter or replace Canon firmware. ML doesn't (re)use Canon IP in its code. Therefore ML is not prone to legal action.
As far as I understand your request you want to replace Canon's firmware with a version where ML features are integrated in firmware. IMO this violates Canon IP. Won't happen here.

eduperez

  • Developer
  • Freshman
  • *****
  • Posts: 92
Re: Canon 70D
« Reply #3080 on: January 17, 2018, 04:52:36 PM »
I was reviewing the list of Magic Lantern features which cannot be implemented on EOS 70D due to firmware limitations.  Okay, I don't know a thing about the CPU on this device or what language its firmware was written in, but I would imagine that somebody had to run a decompiler on the firmware at some point, while verifying what limitations it presents when porting Magic Lantern to run on it.  Furthermore, I would imagine, that if it could be demonstrated that the source code of decompiled firmware was sound enough to be compiled and link-edited to create machine code identical -- or at least functionally identical -- to the original firmware from which that source code was decompiled -- well, then I'd imagine the source code could then be tweaked so that the desired missing features of Magic Lantern would then run on EOS 70D.

Now, I know that's a lot of ifs -- and testing tweaked firmware would not only void warranty but could damage the camera.  But I imagine it could be done -- unless there is an actual hardware limitation on this camera beyond what the firmware will allow.  It might be foolish to try this ... but I cannot help wondering if somebody has?  Anybody in this forum perhaps?

That is not how it works...

Most of the firmware (at least the parts of it being executed on the main CPU) were (most probably) written in C. So far, all decompiling efforts have just translated machine code into assembler, and only for the purpose of studying how does the firmware work; it has never been translated into the original C (except probably some very specific parts, for illustration purposes). Besides, ML does not contain any of Canon's code, neither as binary, source code, or compiled from a decompilaton.

ML does not replace or substitute the original firmware, or even parts of it: what ML does is to interfere in specific points, hijacking the communication between different parts of the firmware, or between the firmware and the hardware; but the original untouched firmware is still in charge of the hardware, most of the time.

What you are proposing is a whole new beast... not only you need to produce a sorce code that can be compiled to match the current firmware, you also have to ensure that any modification you make to it does not break everything. I have still not heard of anyone decompiling the ASM into the original C, or any C code that can be compiled back and produce the same firmware, let alone some C that can be understood by humans. And then, just adding one single instruction means all code and data gets relocated, and thus all references must be updated.