u can kinda see it in the blacks on the table. I actually notice a huge difference. Way less "blocking" for sure.
Magic latern firmware for 5D - 74 MBps h.264 Tests & 2.0.8 Beta Firmware
- Thread starter Retrospective
- Start date
u can kinda see it in the blacks on the table. I actually notice a huge difference. Way less "blocking" for sure.
youngindiefilms
Well-known member
Supposedly Magic Lantern 0.1.6 works with firmware 2.0.8. I copied the ML firmware to the root of my Sandisk Extreme IV (45MB UDMA 16GB), as suggeted in the wiki I turned the auto sensor cleaning off, also using original Canon battery but I can't get ML to work. After I start the upgrade process the camera just locks up, black screen, no reaction. Would love to try the new higher bitrate modes. Any ideas?
wasn't the Q's -9 thru-16 in ML supposed to set it to constant? Or was there something else I missed in the debug menu?
millerandmiller
Member
As the full bit rate causes a lot of buffer stops I thought I would do a test by reducing the visible information on screen with a black card aperture over the sensors OLPF, whilst the shutter is up. Not for the faint hearted as this could screw your shutter if live view fails.
A kind of Heath Robinson 2:35 Ratio.
The result is I can record full bit rate (-16) at 25p. For as long as I need, the buffer only ever reach 1 bar even with lots of whip pans and bright light. Card: 8GB 60MB/s UDMA SanDisk.
Goes without saying, sharpness all the way down in camera.
Even covering the part of the sensor that get cropped on video helps a lot and increases recording time to minutes instead of seconds.
http://yfrog.com/bcyz30j
A kind of Heath Robinson 2:35 Ratio.
The result is I can record full bit rate (-16) at 25p. For as long as I need, the buffer only ever reach 1 bar even with lots of whip pans and bright light. Card: 8GB 60MB/s UDMA SanDisk.
Goes without saying, sharpness all the way down in camera.
Even covering the part of the sensor that get cropped on video helps a lot and increases recording time to minutes instead of seconds.
http://yfrog.com/bcyz30j
Last edited:
millerandmiller
Member
I had the same problem. I went down to 2.0.4 then back to 2.0.8 and it worked with ML.
squig
Veteran
Brilliant!
Do you still get 1920 pixels wide with that black card in place? Another option would be a couple of strips of electrical tape.
youngindiefilms
Well-known member
Got it to work now, thx!... however I can't seem to find how to change the bitrate. Googling it didn't shed any light on this neither is there any info how to activate the higher bitrates on the wiki, unless it's really hidden away somewhere.
Last edited:
millerandmiller
Member
5th Menu across 'Debug' 3rd Item down 'QScale' press down on the joystick to change and cycle through settings. Default is '-8' Highest bitrate setting is '-16' not '+16'.
squig
Veteran
I've been testing a Lexar 600x 16Gb UDMA card, my research suggested that this was the fastest card on the Canon DSLRs but unfortunately it's performing about the same as the Sandisk extreme 60MB/s card. I'm still getting about 60-65Mb/s with the -14 setting. The variable bit-rate is really frustrating especially in low light. I think a constant bit-rate would work a lot better.
millerandmiller
Member
There is space on the left and right of the OLPF, so its possible. Mine has a thin strip left and right, but then it was a very quick cut out. Electrical tape might gam things up as she warms up, also would could leave sticky stuff on my OLPF.
As the top most OLPF is higher than the sensor, the edge of the mask will not be sharp. (have to pull this apart, again)
squig
Veteran
Yeah tape could get messy, now I'm thinking whiteboard marker.
Another possible option could be to get a pro to remove the OLPF and put a mask underneath it. With the MKIII not too far away I probably wouldn't go that far but it's certainly something I'd be interested in exploring to get more bits out of the MKIII. I letterbox all my work in post anyway.
Another possible option could be to get a pro to remove the OLPF and put a mask underneath it. With the MKIII not too far away I probably wouldn't go that far but it's certainly something I'd be interested in exploring to get more bits out of the MKIII. I letterbox all my work in post anyway.
Last edited:
millerandmiller
Member
Fixed rate would be great, should stop the spikes.
millerandmiller
Member
I wonder if the white board marker would be opaque enough and don't they get dusty when you wipe them clean. Mind you could use that as a graduated ND filter.
I think I'm with you on that one, roll on the MKIII.
P.S. you still get vast improvements if you just mask the area of the sensor that gets cropped by standard video mode.
youngindiefilms
Well-known member
I have no QScale option whatsoever. Under the debug menu I get:
CMOS temp
save config
toggle draw_prop
dump dmlog
screenshot
toggle draw-event
millerandmiller
Member
Have you downloaded the correct ML firmware, folder contents look like this.
magiclantern-0.1.9.zip
http://groups.google.com/group/ml-devel/attach/71581e7312e05a2a/magiclantern-0.1.9.zip?part=2
REF:
"I have no QScale option whatsoever. Under the debug menu I get:
CMOS temp
save config
toggle draw_prop
dump dmlog
screenshot
toggle draw-event"
youngindiefilms
Well-known member
That's the one I downloaded. Put the cfg and fir files in the root, obviously it works but there is no qscale option in any of the menus.. also no entries of anything related to the bitrate in the config file when I open it with a text editor.
millerandmiller
Member
There isn't ment to be any reference in the config file, so don't worry about that.
Humm.. very strange.
First thing I would try is to flash everything again. Sorry can't be of more help.
youngindiefilms
Well-known member
I'm totally baffled as well. I definitely got the same version from the link you've posted. All that I put on the cf card (root) was the fir and cfg files plus a custom bmp for the cropmarks. Am I missing anything? Can someone else using 0.1.9 please help?