50D - 14bit raw video builds and test results

Started by GregoryOfManhattan, June 17, 2013, 01:53:22 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

GregoryOfManhattan

hi 1%,
not my own repo - i thought it useful to have a 50D build which tracks exactly what is on hudson unified
so any builds i have called "unified" are from https://bitbucket.org/hudson/magic-lantern/src

towards that end, could you please let a1ex know what you currently are using for skip_left in src/raw.c
https://bitbucket.org/hudson/magic-lantern/pull-request/130/undoing-the-damage/diff

is it useful to continue to have builds that track the main unified?

a1ex

See my comments from https://bitbucket.org/hudson/magic-lantern/commits/5ddc1608bf8fd87137919d7eea7d2627

I did not see any DNGs with left borders, please upload them. If the left offset varies, you can use something that covers all cases, but you have to disable black computation from the left side.

1%

It doesn't change or anything... it is what it was. Only 2 modes & 64/74. Andy's pics were from raw video. IMO, pink /corrupt frames are another problem. You shouldn't get any with GD off @ record... at least I don't.


Ps... lots of cool stuff @

FF833CCC

including lv_hd
lv_debug_ae, etc... lv_fps


LV_ChnageHD OldHD:696 NewHD:1
SetHD 1


That param is wrong but maybe we can biggen or shrink yuv/lv.

Sweet this stops lv.. fps is 60 but I doubt that... maybe it helps write speed or something.


lv_fps does nothing but lv_hd sets native fps.... I've seen 60, no image, 28.. good image... so I have to find a value which will set it to 24 native.

call("lv_hd", 0x3E8); clean 21fps
call("lv_hd", 0x36A); 23.878 fps
call("lv_hd", 0x366); 23.988
so close.

doesn't work in zoom tho :(

lv_magnify ???
lv_defmark_raw ???


defmark does this... called with 1... maybe thats not right, I don't see any less defects.


pfAfComplete(642)
Result = 1
[ENG] WriteDMACInterrupt(17)(0x2)(0)
WriteEDmacQuarkCompleteCBR
[ENG] WriteDMACInterrupt(0)(0x2)(0)
WriteEDmacYuvCompleteCBR
[ENG] ReadDMACInterrupt(9)(0x2)(0)
ReadEDmacHivshd_V_CompleteCBR_x1
[ENG] WriteDMACInterrupt(18)(0x2)(0)
WriteEDmacVramCompleteCBR_x1
SetPsave 0x0
[ENG] ReadDMACInterrupt(8)(0x10)(1)
ReadEDmacDefCorreAbortCBR_x1
[ENG] ReadDMACInterrupt(9)(0x10)(1)
[ENG] DummyCBR
[ENG] ReadDMACInterrupt(8)(0x10)(1)
ReadEDmacDefCorreAbortCBR_x1
StopQuarkPass

Andy600

Quote from: 1% on June 24, 2013, 06:46:45 PM
...Andy's pics were from raw video

The frame grabs from the video were raw video DNGs but a1ex then asked me to send some silent pics too, which I did here: http://www.magiclantern.fm/forum/index.php?topic=5586.msg53730#msg53730

Not that it's relevant now.

BTW, I'm just wondering if some things have not been correctly found for the 50D but instead, 'assumed' from other cameras. There is code commented out to do with raw buffer size in raw.c. It was set the same as the 60d which has a different sensor.

p.s. - interesting finds (above). It's great to know you're looking for this cool stuff 1% :)
Colorist working with Davinci Resolve, Baselight, Nuke, After Effects & Premier Pro. Occasional Sunday afternoon DOP. Developer of Cinelog-C Colorspace Management and LUTs - www.cinelogdcp.com

1%

I've gotten rid of most "assume" stuff. Its actually pretty similar among cameras.

I got LV up to 80fps but image is dead :( But some other numbers make higher fps.. I had working 34fps at one point. Need to dump fps to log file after setting this.. or maybe someone can look.. this could turn on defc on in other cameras, etc.

KENNEDYR21

@GregoryOfManhattan

I just updated to the latest unified build https://bitbucket.org/GregoryOfManhattan/magic-lantern/downloads/magiclantern-2013Jun24.50D.109.go.unified.d07dab0ba6c2.zip

But for some reason with this latest unified build, the modules won't load and I'm getting an Err  next to each module where it's normally and ok.
I haven't had this happen before. Is there a different protocol for this latest unified to run?

GregoryOfManhattan

new hudson unified build posted
https://bitbucket.org/GregoryOfManhattan/magic-lantern/downloads/magiclantern-2013Jun25.50D.109.go.unified.282abfa87fef.zip
quick test shows modules load and recording raw work.

@KENNEDYR21 - sorry about that - module building changed 5 days ago and i mis-adjusted my build script.
it should be fine now and going forward.

these builds should track both the main magic lantern in hudson/unified and in the modules as well.

akumiszcza

In the new build I see speeds required 10x as they should be - 270 MB/s instead of 27 for instance.

KENNEDYR21

@GregoryOfManhattan,
Ok I will update to the new build you have recommended. Thank you very much!

GregoryOfManhattan

Quote from: akumiszcza on June 25, 2013, 07:57:54 PM
In the new build I see speeds required 10x as they should be - 270 MB/s instead of 27 for instance.
good eyes to catch this.
a1ex fixed it - the code for calculating the speed has been changed and may be more accurate now.

1%

Yep, I built after that commit and seems OK.

KENNEDYR21

@GregoryOfManhattan

Thanks for posting the updated build, it worked.

But i have another issue that i have noticed recently. When i connect my Marshall 5inch HDMI  monitor i get the pink corrupt frames, not in the normal video record mode, but as soon as i engage Raw modules.

One of the earlier mid June builds was fine at first but now all of the builds are giving me that same issue.
The strangle thing is once i unplug  the HDMI monitor, the build works fine in Raw record. Do  you or any of the Devs know why is the case and is it a possible fix?
It's such a plus to be able to have the HDMI monitor especially when using my camera rig.
Thanks for your time!


GregoryOfManhattan

https://bitbucket.org/GregoryOfManhattan/magic-lantern/downloads/magiclantern-2013Jun27.50D.109.go.unified.b77a85f7ac68.zip
integrates some of 1% hacks.
runs at 70MB/s without.
enabling "Small Hacks" results in 80MB/s
had 2 shots at 1920x1080p24 go over 1000 frames.

Viente

1%, with your latest build cropmarks not workng while recording, only works in standby mode

1%

With global draw on in the module menu?

Tried, seem to work if I didn't turn off GD while recording.

Viente

Quote from: 1% on June 27, 2013, 03:23:17 PM
With global draw on in the module menu?

Tried, seem to work if I didn't turn off GD while recording.

sorry! my bad...  :-X

Viente


akumiszcza

I just got my KomputerBay 32GB 1000x today and tested it on the newest build - magiclantern-2013Jun27.50D.109.go.unified.b77a85f7ac68.zip

I did not format the card, just copied everything and updated firmware. Things I've noticed:

1. Benchmark test failed at first. It went through till 5/10 and then froze. Even after turning off the camera the test was still appearing.
2. Later benchmark showed:

3. 1584x1058 @ 23.952 at 1x - continuous, 1920x1080 @ 23.952  at 5x - about 600 frames (I did not change RAW to JPEG and probably some other options were not set properly though).
4. When I half-pressed shutter during 5x zoom (for IS to work) the screen turned pink again from gray (the option in movie menu to automatically half-press shutter does not work BTW).
5. While previewing the raw movie I got Footer format mismatch, but the preview worked.
6. When I chose lower resolution for 5x mode (1700-something 16:9) which was continuous on my card, the recording stopped after about 100 frames or so twice. Other tests with the same settings worked OK. Some random thing. The buffers went straight to almost full in a second or so.
7. A file was split in two even though it was very small one - RAW is 78.5MB, R00 is 103.7MB.

a1ex

Quote7. A file was split in two even though it was very small one - RAW is 78.5MB, R00 is 103.7MB.

This may indicate problems with the card. Splitting only happens when a file write call fails.

1%

QuoteBenchmark test failed at first.

On 50D mine fails on the read tests in LV and camera locks up... suggests that its a repo problem tho, not something I did.

araucaria

I just tried magiclantern-2013Jun27.50D.109.go.unified.b77a85f7ac68 with a Komputerbay 32 1000x in 1920x1080 crop mode, got over 1100 frames and stopped it because it was reaching 4gb. It said 2000expected and it kept growing, the buffer graph was only slowly filling.

GregoryOfManhattan

@akumiszcza thank you for your continued detailed bug reports - how did it work with a card format?

@araucaria are you saying that spanning at 4GB is not working for you.

@1% failing read test in LV is not a good sign is it?

starting the Card Benchmark while in LV gets substantially slower results for me.

turning on camera and running benchmark got these results



turning on camera, turning on LV and then running benchmarks got the following



araucaria

Never tried file spanning, 300-400 frames is already too much for the things I shoot, and I don't want to waste write cycles on my card, that's why I didn't go any further. Great job btw!

GregoryOfManhattan

https://bitbucket.org/GregoryOfManhattan/magic-lantern/downloads/magiclantern-2013Jun28.50D.109.go.unified.3ecd254bdba1.zip
new "Warm up" the card option for raw record to potentially help the speed of the first recording.

Note: last build from me for a while - i will have limited internet for a couple of weeks.