Messages

Thanks for this, I've been getting quite frustrated by the pink! I'll give this a go tonight 

Is there a way to set AF focus at infinity / nearest?

I'm trying to write a script that takes several photos at different AFMA values, to load into Reikan. Before it takes each photo I want to set focus to infinity.

Have you tried using something like ffmpeg to create an image sequence from a video, and then scrolling through the images to see if the motion is smooth? it could just be your computer struggling to keep up

I'm wondering: how is the RAW data (on the camera, before it is copied) aligned? is it 14bits with two zeros packed in 16 bits, or is each pixel packed tightly together?

Wow . . . wait a minute, Would I be too presumptuous to assume that if this works on the 5D2, this may very well work on the 50D!? Remember, the advantage of the 50D is that it takes CF cards - including the 1000x and has less overhead without audio . . .

Please say it will, Please say it will, Please say it will . . .

raw pictures work for 50D, I posted an example earlier in the thread, but burst pictures will need CONFIG_FULL_EXMEM_SUPPORT for 50D, which isn't yet available. I started adding support for 50D in raw.c but I've not found time to work out the correct offsets etc yet, hopefully I'll manage that in the next few days (or someone will beat me to it!).

Is it possible to directly access the storage (CF/SD) at the block level? I'm thinking that for cameras with dual card slots, an unformatted (hopefully this would stop other canon code accessing it) card could be used to directly store the raw LV data without a filesystem. This might be a wild idea, and I have no knowledge of how the DMA works on the canon chips, but other ARM devices allow you to do DMA from one peripheral to another without involving the cpu - could we pipe the raw LV data directly to the CF/SD periphal, writing a block at a time and some sort of basic linear file system?

The 5D2 address / settings works for 50D. Pictures are 1664x1164. Sample Image

How is the RAW_LV_EDMAC address found? I opened the EDMAC tool on my 5D3 but none of the addresses it finds seem similar to the one in consts.h.

silent_pic_take_raw() calls 'lv_save_raw', is there a way to find all callable functions (I assume this is a canon function?) from a firmware dump?

I'd like to help out with ML dev so I'm trying to understand what is going on here  maybe there are some experiments I can run on my 50D to see if that can support raw silent pics?

If I set the temperature, green and exposure to match yours I get the same image, so it must be guessing them wrongly for some reason. Very strange!

It opens just fine with ufraw here.

I just tried with ufraw 0.19 and got the same result:

I just tried it out on my 5D3, but it is corrupted in everything I've tried to open it with - Shotwell, Darktable, gimp, gThumb.

here's an example file: https://www.dropbox.com/s/5c2vah29wis0wjk/52090007.DNG

Awesome, means I don't have to drag my 50D around for timelapses so that I don't wear out my 5D3 

Great job!

I've tried testing this on my 50D, but I'm not sure what you mean by 'set the screen to Dim and Turn off'. Is this a Canon or ML setting?

I'm thinking about experimenting with ML dev, and seeing this ability to recover from problems like this is encouraging! if you mess up properties is it generally still possible to boot the autoexec.bin and (manually) fix stuff?

Is there a way to dump all properties (from a healthy camera), making a backup that could be used in case of disaster?

I've seen these sorts of weird compiler errors before when doing embedded arm development, generally they were fixed by changing toolchain. It looks like you're using a Mac so I'd try this toolchain - just extract somewhere and change the path in Makefile.user and see if it helps. If it's a clean checkout then the compiler looks a likely candidate.

Another option is to detect the sync pulse on the analogue video out (e.g using an LM1881). This will probably lag behind the actual exposure, but hopefully by a constant amount that could be characterised for each camera.

It's not on the nightly build, but if you check out the latest code and build it, you will get the intervalometer. To run the autoexec.bin you need to enable the bootflag using the special .fir from the first post, and make your card bootable.

50D worked without much effort - just copied 60D definitions, changed to 50D and added the appropriate run_ml script.

I tried to generate a patch for the changes but hg diff was producing a lot of spurious changes - maybe it doesn't like diffing a diff?!

Sorry, I didn't read that well! I thought I had the 5D3-ROM in the correct location but I guess not. I now get as far as create_init_task so I assume that qemu is at least working. I'll have a go at adding support for 50D.

I'm trying to run this on Arch linux:
I had to build with these options because python 2 binary is called python2 and the docs building didn't work:

Code Select Expand

./configure --target-list=arm-softmmu --python=/usr/bin/python2 --disable-docs

when I try to launch:

Code Select Expand

➜  qemu  ./run_ml_5D3.sh
make: Entering directory `/home/tom/dev/software/qemu/qemu-1.4.0'
make: Leaving directory `/home/tom/dev/software/qemu/qemu-1.4.0'
make: Entering directory `/home/tom/dev/software/magic-lantern/platform/5D3.113'
[ VERSION  ]   ../../platform/5D3.113/version.c
[ CC       ]   version.o
[ MENU IDX ]   menuindexentries.h
No menuindex.txt not running "python2 menuindex.py"
[ CC       ]   menuindex.o
[ LD       ]   magiclantern
[ OBJCOPY  ]   magiclantern.bin
[ SYMBOLS  ]   magiclantern.sym
[ CC       ]   reboot.o
[ LD       ]   autoexec
autoexec.bin: 443984 bytes

Program Headers:
  Type           Offset   VirtAddr   PhysAddr   FileSiz MemSiz  Flg Align
  EXIDX          0x06c550 0x000d5950 0x000d5950 0x00008 0x00008 R   0x4
  LOAD           0x000100 0x00069500 0x00069500 0x6c458 0x7bb6d RWE 0x100
[ OBJCOPY  ]   autoexec.bin
make: Leaving directory `/home/tom/dev/software/magic-lantern/platform/5D3.113'
make: Entering directory `/home/tom/dev/software/magic-lantern/platform/5D3.113'
make: `qemu-helper.bin' is up to date.
make: Leaving directory `/home/tom/dev/software/magic-lantern/platform/5D3.113'
rm: cannot remove 'vram.txt': No such file or directory
rm: cannot remove 'vram.png': No such file or directory
00000000 - 3FFFFFFF: eos.ram
40000000 - 7FFFFFFF: eos.ram_uncached
F0000000 - F0FFFFFF: eos.rom0
F1000000 - F1FFFFFF: eos.rom0_mirror_F1
F2000000 - F2FFFFFF: eos.rom0_mirror_F2
F3000000 - F3FFFFFF: eos.rom0_mirror_F3
F4000000 - F4FFFFFF: eos.rom0_mirror_F4
F5000000 - F5FFFFFF: eos.rom0_mirror_F5
F6000000 - F6FFFFFF: eos.rom0_mirror_F6
F7000000 - F7FFFFFF: eos.rom0_mirror_F7
F8000000 - F8FFFFFF: eos.rom1
F9000000 - F9FFFFFF: eos.rom1_mirror_F9
FA000000 - FAFFFFFF: eos.rom1_mirror_FA
FB000000 - FBFFFFFF: eos.rom1_mirror_FB
FC000000 - FCFFFFFF: eos.rom1_mirror_FC
FD000000 - FDFFFFFF: eos.rom1_mirror_FD
FE000000 - FEFFFFFF: eos.rom1_mirror_FE
FF000000 - FFFFFFFF: eos.rom1_mirror_FF
C0000000 - CFFFFFFF: eos.iomem
eos_load_image: file not found 'ROM-5D3.BIN'
run_ml.sh: line 15: 25600 Aborted                 (core dumped) $QEMU_PATH/arm-softmmu/qemu-system-arm -M ML-$1
convert: unable to open image `vram.txt': No such file or directory @ error/blob.c/OpenBlob/2641.
convert: no images defined `vram.png' @ error/convert.c/ConvertImageCommand/3103.


My card was full, formatting did work, and the shoot_malloc contig went up to 26mb. I restarted the camera and it is back at 3mb.

Here's the memory screen for 50D

I'm not sure if the values are correct, it is giving me quite a large adjustment value (-12). Should I have just one focus point selected?

ah ha, you're right, there was not enough contrast!

it works now 

If you change AFMA value from ML menu, does the value get updated in Canon menu? Try in both modes (by-lens or all lenses).

It does; I can set the 'all-lenses' value to 2 in ML, it is set as 2 in Canon. I change again in Canon menu, it is reflected in ML menu.

I can't see how to change between by-lens and all-lenses in the ML menu however?

Any idea why this should be?

you need to checkout and build the latest code