Personal View site logo
Raw video on Canon 5d Mark 3 and other cameras using liveview framebuffer
  • 547 Replies sorted by
  • LOL@maxr. You're right dude, the first vid didn't look so good second (sober) viewing haha. Love the gh2/3s and wanna see this booster/turbo lens at work with them!

  • @driftwood

    can you explain (or post a link) the "set ETTR"? I didn't see an explanation over at magiclantern.

    thanks

  • ETTR = "Expose to the Right". Google that and you're good. Cheers!

  • @roccoforte I dig your 50d footage the most. It had its own look that didn't look like everyone else's ML Raw. Well done :)

  • hi @driftwood. what does the extra buffer mean? what is the benefit? thanks

  • @Fulgencio 2 x 32M extra buffers = Performance increase. Still being merited as we speak but the June 9th build is very stable now. Busy now sorting out the h264 encoder on the mkiii, I wanna bring 'moon' style 150M+ resources with matrices to its rather soft defaults. Also a chance to open up the 8x8 adaptive transform would be cool.

  • i've heard a lot about aliasing with the 50D.. any examples of it? i've seen moire but not so much aliasing that I can find.. some images are noisy however

  • Building edge in the foreground.

  • Interesting

  • Komputerbay cards

    I ebayed up a 64GB Komputerbay card and tried it with RAW. I got errors writing end of data, even after formatting the card in the camera.

    I reformatted and partitioned it on my Mac as ExFAT rather than FAT32.

    It is now behaving.

    Wrangling

    All this wrangling is interesting, I have available: GoPro3, GH2 Sedna'd up, a BMCC and a 5D3. It would depend on what I was doing as to what I would use. My old matey @anthonybert was round the other day teaching me about cinematic (rather than crispy video grades). We were looking over previous projects for bits of footage to grade. I found some footage that I swore blind was 5D3, turned out it was GH2 Sedna.

    For me the 5D3 stuff is the best I can get, followed by BMCC then GH2, but I can be wrong!

  • Latest ML nightly raw build (13th june build) seems to have cleared up the artefacting to my eyes - best workflow: RAW2DNG import/interp to Adobe After Effects, comp it and render out to Go Pro CinemaRAW and open it in Resolve for grading. Love Resolve on the DELL 2913WM.

    DavinciResolve.png
    2560 x 1078 - 1M
  • Who's that swarthy looking guy? :)

  • Here's a little quick comparison I did GH2 vs. 5D MKIII raw. Just two shots. 5D looks great, but surprisingly GH2 also not too bad in this extreme scenario:

    I recently got very interested in the RAW capabilities of the 5D MKIII hack as a possible upgrade path from my trusty hacked GH2 / GH1 combo. Hence, I asked my friend Peter Hainzl (have a look at his RAW videos) to come over for a quick testshot during lunchbreak. As my hacked GH's are fantastic all around cameras, tack sharp, almost moire free and perfect for shooting as they are out of the box (without additional rigs / whatever) but lacking a bit in the dynamic range and lowlight department, plus the limitation of 8 bit 4:2:0 color space I was wondering if the 5D MKIII would fit my bill. I am mostly interested in landscape / nature / people, hence I chose a very high dynamic range landscape scenario for this quick test - shadows from trees and bushes vs. sun - illuminated clouds in the skies. Hence, time for pixel - peeping (yes, about every 2 years is OK for me ;-)).

    follow me on twitter: twitter.com/gunmac

    Setup: GH2 with Voigtlaender 25mm f0.95 at f2, LCW vari - ND, smooth film mode (-2,-2,0,-2) mounted on top of 5DMKIII with Canon 50mm f1.4 set to f=4 (to get similar DOF characteristics to Voigty at f=2) with Heliopan Fader ND. Both at 1080p23.976. Same scene recorded with both cams. You will notice a slightly wider field of view on the GH2, as it has a crop of about 1.86 to the 5D MKIII.

    Workflow: RAW images debayered in Adobe After Effects CS6 with Adobe camera raw, saved as composition. Imported AE composition in Premiere Pro CS6, sequence set to max. bit depth. Rendered 1080p file as 25mbits/s H264 file for vimeo upload.

    Findings: - sharpness / resolution detail especially in shadows of RAW 5D MKIII better than GH2 - better dynamic range, especially in highlights, better details in shadows BUT surprisingly not so much more. In this scene I expected blown out skies with the GH2 (see first scene), but surprisingly could pull back a lot of the highlights as well as the shadows - Shadow noise is severe for both cams, could be improved with 5DMKIII by exposing more to the right, as there was still room in highlights - in general much cleaner image, nicer gradations with RAW 5DMKIII - not very surprising as we are talking 8bit 4:2:0 vs. 14bit 4:4:4

    In summary I am very happy I did this first quick test, as it revealed a lot to me - I am mainly interested in internet distribution of my films. And for this purpose the images are NOT lightyears apart. Of course the image is lovely lovely lovely on the 5D. But for travelling, it is also much much bulkier and the screen is fixed which is difficult for my filming style. With careful exposure on the GH2 you can still get decent pictures. Next thing I will try is the blackmagic pocket cinema camera. I ordered it on the first day, so lets see if I get it sooner or later. I am still interested in raw, as the image manipulation possibilities (color grading, exposure, ...) on the raw files are sensational.

    Download the source file and have fun!

  • No offense dude ----because I KNOW you worked hard on this but if you are gonna compare the 2 cameras use a better patch like Intravenus V2 not a weak 45MB patch...its really kinda a pointless and useless comparison with such a LOW bitrate patch- Sorry not trying to be a dick but if your gonna compare Gh2 to the Framebuffer raw Mk3 you should be using the BEST Gh2 has to offer--also no offense to the makers of the lower bit rate patch which have their uses.

  • @No_SuRReNDeR, A better patch isn't really going to improve the results of a dynamic range test that much. Maybe in the macroblocking of extreme shadows, but not the clipping by much.

  • I know it doesnt affect DR that much but its just retarded IMO to not compare BEST with BEST....

  • Nice test GMC, thank you! I was surprised how our wonderful little GH2 looked compared to mk3 raw - it was inferior, obviously, but the difference wasn't miles apart. Yeah, you could've (and should've) used a better patch, but imho it wouldn't make a huge overall difference.

  • @mordae just replied faster than me, and he is totally right - @no_ surrender I am happy to accept any explanation on how a higher bit rate hack would improve dynamic range. Also, if you read through my summary you will find that I was quite surprised that the final result did not look too different. Which basically cooled down my "raw" excitement quite significantly....

  • Its a nice test please dont get me wrong... just wished you would have used a better patch is all I am saying-and it does make a slight difference in shadows etc...

  • @no_surrender I understand your point regarding shadow macro blocking. However that was not the point of this test, and I guess that's clear by now.

  • One thing this shows is if the GH-2/3 ever gets a RAW hack it would be that much better. Right now it's holding its own in DR and sharpness against a RAW FF camera. Unbelievable. ....and here I was thinking of pulling the trigger on another Canon camera because of the RAW. This actually stopped me in my tracks. Makes me anticipate the coming on the BMPC's wider DR even more. Thanks @GMC.

  • @Ian_T I agree with you, but not on the raw GH2 part. What I have seen comparing video sequences and raw photos of the same scene with the GH2, that there is a little additional info in the raw photo, but not very much. The video implementation in the GH2 is obviously really good.