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 - kitor

#1
Camera-specific Development / Re: Canon EOS R / RP
March 14, 2024, 11:59:15 AM
QuoteDo you know how Canon limits the focus on RP at 4K? Wondering if it's a Hardware or Software limitation.
No idea, I don't have a working RP, nor I investigated focus stuff.

QuoteWith the fast card you have, recording at 3k@24, do you see any temp bump, beyond the one you posted before?

Those were like 3 minutes bursts, I didn't see any high temp rise. Hard to test with so small card for the job though :)

#2
D8 cams have buffer >= 1920x1080, so is is true 1080p+.
Not tested for line skipping/binning atm (R does binning).
#3
How low spec we can get with mlv_lite?

SX740 the answer... at least for now :)
Yes, the jpeg-only cam now records RAW too.


Interface seems to max out at 86R/80W. Not enough for continuous 1080p at a normal frame rate.
But tbh that was mostly a test "if sx740 can do it, every model can do it".

SX740 lacks EDMAC mem2mem copy code which I partially used on R so I had to reimplement whole EDMAC memcopy using pure EDMAC functions and cross my fingers that channels used by R were unused here. And it worked fine :)

#4
8.8.1 / 8.8.4 is internal firmware version.

We can't help with 1 series as we don't touch those. As written above, Tornado EOS will likely help you.
#5
Camera-specific Development / Re: Canon EOS R / RP
February 26, 2024, 11:09:55 AM
Some fun tests with fast card (Adata 64GB 290MB/s).
Benchmarks top in-cam at 220MB/s write. That's enough for continuous 1080p60 and borderline for 3k@24p (literally, requires card warmup to not stop immediately).

This also means 4k 16 bit uncompressed is impossible at usable FPS. Maybe 10 bit lossless, but at this moment we don't know how to do 10 bit or how to do lossless.
#6
g3gg0 restored wiki from backup, as missing SMF integration led into massive spam infestation. It is integrated with SMF again (requires separate login, but with forum credentials)
#7
Videos, both youtube and vimeo should be fixed. Took a plugin... and some fixes to that plugin.
If any is still not playing (but if you use qoute to get the link and it plays fine directly on vimeo/yt), please link it in this topic.

Also InfoCenter shows again just the last message in given topic.
#8
All video embeds are broken. Turns out prior to now SMF had to use a plugin, now they support the same tags out of the box... but changed how they work because why not.

There's a workaround posted on their forums... hopefully it will fix them without editing all posts in database.
#9
One more: youtube embeds are broken.
#10
Quote from: reddeercity on February 22, 2024, 12:58:05 AMSo what happen here ? I'm lost can't even found the latest posts it's so out of place even going to desktop version doesn't help . I hope the old site is coming back soon.

SMF (forum script) was updated to a new major version. Not sure if old style (which was default SMF style for years) is even included with the new one.

Quote from: Walter Schulz on February 22, 2024, 01:44:36 AMFormer version: Only *latest* post in a thread.

Ah yes, that felt odd while beta testing update yesterday - but I didn't catch what exactly has changed, only that something did.

Quote from: reddeercity on February 22, 2024, 12:58:05 AMAnd what with the "I agree" thing when I logged in ? It doesn't even give the terms of the agreement so what I'm I Agreeing too here ?
Already gone or I'm not sure where it is?
#11
Camera-specific Development / Re: Canon EOS R / RP
February 19, 2024, 09:12:07 AM
Ok, forgot about that one  8) Nah, SX740 - it will be fun to see if this point-and-shoot can do 1080p30. D8, gig of RAM, only UHS1 though. Seems enough.
#12
What is your distro? Old Ubuntu required for old repo?

permission denied is weird, try to run that script directly, maybe you are missing python3 or git. Or somehow it is missing executable bit.
This is literally a python3 script that runs two git commands.

kitor@kitor-p73:5D2.212$ ../../build_tools/git_summary.py
268224668 raw_draft_v2%
#13
Camera-specific Development / Re: Canon EOS R / RP
February 19, 2024, 08:42:28 AM
Linked attachment requires logging in.

Quote from: Danne on February 18, 2024, 09:23:50 PM
I so regret selling my eos  rp now.

I kinda expect RP to be the same +/- sensor.
Can't test as I'm still hunting for a donor RP and 250D motherboards - RP was sent my way by one of our community members; 250D i found myself - both exhibit the same issue with power circuit. I tried reviving those, with a great help from @vth (from photo-parts.com.ua) but it seems I got unlucky and got too many dead PMICs (which are unavailable except sourcing them from Canon boards). Funny is that IIRC @coon has two dead RP boards that exhibit similar issue - only MPU wakes up and complains about PMICs i2c responses.

Anyway, I'm going to check another D8 cam soon. The most ridiculous one :)
#14
Camera-specific Development / Re: Canon EOS R / RP
February 18, 2024, 09:11:42 PM
I found the code, R has 5 sensors: SH, MAIN, A, WM, BACK.
I did a test, started with 27, 32, 31, 29, 26 C. Recorded half of 64G card in 1080p30 (that was 3 minutes). Temps after recording: 29, 35, 33, 29, 27.
#15
Camera-specific Development / Re: Canon EOS R / RP
February 18, 2024, 02:10:54 PM
Not tested, tbh 64 gig card I have is good for what, like 5 minutes of recording?
And there's no temperature reading integrated with code right now, I had some PoC stashed (works differently than old models)
#16
Camera-specific Development / Re: Canon EOS R / RP
February 17, 2024, 10:25:57 PM


Turns out I just forgot I have two nearly identical cards: Lexar 64GB x1000 and x1666. And I used the slower one.
With x1666 (~100MB/s write) I get continuous 1080p30 14 bit RAW.
#17
A very RAW update, February 2024

Another small, a bit delayed (for a reason), but this time a very exciting update!

To put this in context: Draft for Q4 update was ready for over a month now. But the situation was so dynamic we decided to push it for a bit later.
As you will see, this was definitely worth the wait :)

Chapter 1: RAW video on 7D2 (Digic 6)!

To the point: @heder recorded a first clip of RAW video on 7D2!

No builds, no code - don't expect them soon. It was a very dirty PoC. Proves it is possible though :)

Quote from: heder on January 17, 2024, 10:53:04 PM
I have RAW video up and running. (...) But going there without ML RAW video is bad karma, so I had to investigate every possible way of getting the camera to do as I wish.
(...)
Im was recording the entire RAW buffer including black border, 1984x1068x14bits,  around 85 MiB/s for 24 fps. I will try to record a large file using a fast CF card ( DMA-6)  in the following days to see if I get frame drops. The following video was recording using 25 fps, and recoded to H.264





Chapter 2: RAW video / mlv_lite on 200D (Digic 7)!!

@names_are_hard had a breakthrough in understanding new EDMAC API on Digic 7 models.
With that, after a bit of hacking now 200D records RAW

Quote from: names_are_hard on February 09, 2024, 04:08:05 AM
Several years of learning ML code, trying to understand fragmentary docs, and in some cases writing the docs myself, 200D has raw video:



At the moment this is only 14 bit, no lossless mode, with 40MB/s limit of SD card interface it is enough for continuous 720P recording.

There are no builds available for now, only dirty source code. For more details visit the quoted thread.



Chapter 3: RAW video / mlv_lite on EOS R (Digic 8 )!!!

Remember how I always written "don't count on D8 models soon?" This was because D8 models changed EDMAC once again, only this time also hardware has changed.
And we found no patents to put those new things into any perspective.

Yet, after @names_are_hard got 200D to work he pointed me to some bits of R code that may be useful. And they were.
We don't understand new controller very well, but it was enough to make it work for mlv_lite. Which after a few days of hard work turned into this:

Quote from: kitor on February 11, 2024, 05:50:20 PM


Yes, that's EOS R. Yes, it is Digic 8. Yes, we said don't count on it as EDMAC is completely different.
Yet here we are  8)

Again, no builds yet. This one is ever more hacked on top of what Names did for 200D. At this moment whatever resolution and frame rate I choose, R is able to get me around 30-60 frames of RAW video.
This is not a limitation of SD interface (it is SDXC, my card easily benchmarks over 100MB/s write), but rather some bug in a code I didn't trace yet.

In any way, we made a significant process in understanding how EDMAC works in "new world", you can expect technical documentation for those coming somewhere in a near future :)

Update:
Quote from: kitor on February 17, 2024, 10:25:57 PM


Turns out I just forgot I have two nearly identical cards: Lexar 64GB x1000 and x1666. And I used the slower one.
With x1666 (~100MB/s write) I get continuous 1080p30 14 bit RAW.

Chapter 4: What can I expect from you? How can I help?

As always, a gentle reminder that ML project is not a company or commercial institution.
This means that while we are happy to listen to your suggestions and feedback, the project doesn't have any timelines, obligations and we are not bound to fulfill any of those.
If there's a build available for your camera - this is all we can offer. It includes all the features and also bugs you may encounter. Everything else should be considered as our best effort to support the community.

What is the status for X? When you will implement Y? Why not Z?
We realise people ask these questions because they're enthusiastic about ML - people want it on their cam and they want it to work well!  This is understandable.
When you have a hundred people asking the same questions, it can come across as nagging, or demanding.
Please try to remember that we will always announce significant progress on the forums, and we will always ask for testers if we need them.

For people who have technical knowledge and want to contribute - there's source code, QEMU emulation, Wiki, development Discord and many more.
We are aware our build system is very outdated and documentation is very lacking. Please consider joining Discord server - we will be more than happy to help you get started.

What about ChatGPT?
Slightly out of context, but since this question is asked very often in every single open source project on the planet.
ChatGPT will not help Magic Lantern. The majority of the work is in reverse engineering (and in general research). AI is trained on already available data - it can help you with problems that were already solved elsewhere.
#18
Camera-specific Development / Re: Canon EOS R / RP
February 11, 2024, 05:56:01 PM
(X-post from 200D shoots raw video)

Quote from: kitor on February 11, 2024, 05:50:20 PM


Yes, that's EOS R. Yes, it is Digic 8. Yes, we said it don't count on it as EDMAC is completely different.
Yet here we are  8)

Caveat: it only adds a page into a list of digic 8 instabilities with ML, records roughly 30-60 frames before stops... something related to memory management that I didn't solve yet.
No builds will follow as D8 requires a ton of work before being usable in any way.

But: AF works while recording  :) When 4k is selected in Canon menus, raw buffer is 4K too (unfortunately cropped, as expected).

My PoC on top of @names_are_hard PoC is on my branch: https://github.com/kitor/magiclantern_simplified/commits/raw_draft/
This is shameless "fake it till you make it", thus needs to be redone from scratch after fixing stuff.


There's also a significant progress in understanding D8 EDMAC. Not enough to setup random transfers, but it was more than good enough to get mlv_lite working :)
I'll wait for Names for his EDMAC documentation and update it by my current D8 knowledge.
#19
Camera-specific Development / Re: 200D shoots raw video
February 11, 2024, 05:50:20 PM


Yes, that's EOS R. Yes, it is Digic 8. Yes, we said don't count on it as EDMAC is completely different.
Yet here we are  8)

Caveat: it only adds a page into a list of digic 8 instabilities with ML, records roughly 30-60 frames before stops... something related to memory management that I didn't solve yet.
No builds will follow as D8 requires a ton of work before being usable in any way.

But: AF works while recording  :) When 4k is selected in Canon menus, raw buffer is 4K too (unfortunately cropped, as expected).

My PoC on top of @names_are_hard PoC is on my branch: https://github.com/kitor/magiclantern_simplified/commits/raw_draft/
This is shameless "fake it till you make it", thus needs to be redone from scratch after fixing stuff.


There's also a significant progress in understanding D8 EDMAC. Not enough to setup random transfers, but it was more than good enough to get mlv_lite working :)
I'll wait for Names for his EDMAC documentation and update it by my current D8 knowledge.
#20
Camera-specific Development / Re: 200D shoots raw video
February 10, 2024, 10:08:54 AM
Quote from: names_are_hard on February 09, 2024, 06:53:20 PM
Zoom res check will have to wait, sorry - cam not connected now, and I'm going to be working on documenting EDMAC stuff for the next few days, so the next dev doesn't have to suffer as much.

Where, by the way, as for yesterday's evening:

     3389:  73978.898 dst, src: 0x4012a000, 0x4014a000
     3390:  73978.905 Pre-copy, *dst, *src: 0x1111, 0xa5a53333
     3403:  73999.170 Post-copy, *dst, *src: 0xa5a53333, 0xa5a53333


That was EOS R EDMAC memcpy. It seems that we'll be able to use it for simple mem-to-mem transfers even without knowledge about all the extra stuff that changed in D8 (and we have no idea how this really works now due to all extra layers).
#21
Forum and Website / Re: Image linking test
February 05, 2024, 03:39:58 PM
Imgur blocks hotlinking for a good year or so now.
#22
Camera-specific Development / Re: Canon 7D Mark II
January 25, 2024, 08:37:34 AM
And btw, DIGIC 8 EDMAC MMIO registers doesn't match old ones at all (offsets, number of registers used for various stuff). That was expected after we found it uses extra dimension... but even if we understand the new controller, integrating that without structures would be a total mess.

Btw, @names_hare_hard I think we should probably pull this recent discussion into a separate thread.
#23
Camera-specific Development / Re: Canon 7D Mark II
January 24, 2024, 09:31:08 AM
Yup, @names_are_hard wrote unpacking again in C... and it works, both for 200D and R.
R turned out to be 2304x922 (including black borders), 2257x900 when cut. So many bad pixels ::)
#24
Camera-specific Development / Re: Canon 7D Mark II
January 23, 2024, 09:46:09 PM
On R (D8) this evproc sequence saves the raw buffer on card. Format unknown, selected by lv_set_raw_wp.  There are 26 formats (?) to choose from, coming from SantaPath or SAP Path. edmac_info source suggests biggest size is 2016x921 (?) which when decoded as YUV produces completely broken image with somehow visible outline of original context.


lv_set_mm 1
lv_set_raw_wp 0
lv_save_raw 1
lv_raw_dump


Reference frame


Source file (buffer dump): https://kitor.pl/eos/r_edmac/test1.raw

Source decoded as YUV
#25
Camera-specific Development / Re: Canon 7D Mark II
December 20, 2023, 10:53:16 PM
Quote from: heder on December 20, 2023, 04:52:44 PM
Thia is really convenient and will make the port more easy, but still it's a strange beast, "dual core + single core"

Huh? D6 is one R4 core (Marius, what we call ICU) and one secondary R4 (Omar), also one secondary Xtensa (Zico, GPU accel) and two tiny cores (Arima, Shirahama; IIRC Xtensa, in single CPU models related to lens comm. They run tiny dryos variant like MPU)
7D2 has two physical D6 chips, thus twice that. So four R4 cores in total across two physical CPUs.