Mis-information by the pros about ML raw video

Started by Andy600, September 16, 2013, 04:02:21 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Andy600

I've noticed a couple of blog articles from professional shooters recently that seem to contain mis-information about ML raw video.

In Nino Leitner's article http://www.ninofilm.net/blog/2013/09/16/kineraw-mini-vs-5d-mark-iii-raw-dynamic-range-test/?utm_source=feedly (posted today) he suggests "It seems like you can "bake in" a picture profile look into the 5D3 raw files."

and in videos that accompany an article by Paul Kell (also retweeted by @autoexec.bin on the front page of Magic Lantern) http://www.eattheblinds.com/2013/09/once-you-go-raw.html he clearly states that QScale VBR -16 has an effect on raw video aliasing.

Although I think these quotes are totally incorrect, what would make a professional come to these conclusions? Are they simply mistaken or is there any substance to what they are suggesting?
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

Midphase

To me, the Magic Lantern raw development is a major game changer, almost as much as  the ability for the 5D2 to shoot video in the first place. It elevates the video quality of these cameras to the levels of RED and Alexa while adding virtually nothing to the cost besides faster CF cards.

So I'm left wondering why either the complete lack of interest on the part of the official press, or at most a superficial interest ripe with mistakes and misinformation? Digital Video Magazine has yet to make any mention of this, Phil Bloom seems to have made a minor acknowledgment and then never mentioned it again, nothing really from the likes of Vincent Laforet or Shane Hurlbut who up until now have been champions of DSLR shooting.

Yeah....head scratcher!

Andy600

Maybe because raw video is still Alpha but also, Philip Bloom and Vincent Laforet have relationships with Canon which must prevent them officially or unofficially being seen to support ML plus neither uses lower end DSLRs anymore. They also have a large number of followers and I guess they wold be concerned about recommending what THEY see as a 'hack' to their followers because they would be in the firing line if something went wrong. Many users live by what these guys say and don't read the forums which is a shame, because most of us know how stable and useable raw video is.

I would be surprised if Shane Hulbert doesn't jump onboard sooner or later, even if it's just for his crash-cams.
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

mannfilm

Back in the '90's, the establishment with their $50,000 Betacam SP camcorders, insisted that DV could NEVER be broadcast quality. They were lying. The history of video is one of the establishment repressing or bad mouthing revolutionary new gear and processes to protect their own gear investment and discourage competition. 10 years ago, the pro's would have been spreading lies about how ML bricked your camera. Today, if they tried it, social media would reveal them and ruin their reputation. So instead they have chosen quiet repression, along with hard to prove or understand allegations.  If Bloom came out for ML RAW, his corporate sponsors would dump him, and his "pro" following, desperate to protect their equipment investment and scare off potential competition, would leave him.

Andy600

In Philip Bloom's defence, he is currently hosting 3x video tutorials on his website by David Kong who is a staunch supporter/user of Magic Lantern.

I think PB personally has no need for ML. He no longer uses cameras that have ML and I think his only Canon in regular use is a 1DC which is top-end.

These guys are simply moving up and I have no problem with that. They did a lot to grow the DSLR video community and now, 4-5 years on they are pursuing their own goals.

There are still plenty of gurus out there for indie DSLR shooters and room for more ;)
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

Africashot

A comment I made on Philip Bloom's website comparing the early ML RAW workflow with what I imagine must have been like working with celluloid film on an article about shooting discipline never got approved, not sure why but I found it rather strange...

Also on other sites I keep hearing claims about overheating, or people saying they will never use a hacked camera for a 'professional' project but then seem to have no problem using consumer lever dslr cameras to shoot those same 'professional' projects...
ML 5D2 & T3i

Andy600

Quote from: Africashot on September 16, 2013, 08:12:02 PM
A comment I made on Philip Bloom's website comparing the early ML RAW workflow with what I imagine must have been like working with celluloid film on an article about shooting discipline never got approved, not sure why but I found it rather strange...

Also on other sites I keep hearing claims about overheating, or people saying they will never use a hacked camera for a 'professional' project but then seem to have no problem using consumer lever dslr cameras to shoot those same 'professional' projects...

He probably saw your comment as the can opener to a raw can of worms :D
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

David J. Fulde

Quote from: Africashot on September 16, 2013, 08:12:02 PM
A comment I made on Philip Bloom's website comparing the early ML RAW workflow with what I imagine must have been like working with celluloid film on an article about shooting discipline never got approved, not sure why but I found it rather strange...

Also on other sites I keep hearing claims about overheating, or people saying they will never use a hacked camera for a 'professional' project but then seem to have no problem using consumer lever dslr cameras to shoot those same 'professional' projects...

Yet they have no problem using a hacked GH2/3 >_>

1%

Just because they're pros doesn't make them computer literate. I think people try things, they work for them but they don't realize its placebo.

Quote
Yet they have no problem using a hacked GH2/3 >_>

GH2 is no danger to the industry, just like hacked H264 here wasn't. Nothing in this price range compares to the raw... it beats 4:2:2 pro cams that were $5k+... maybe not such a jump for 5DIII but the lower tier CF cameras, you have a red for <$1000 and a little bit of hassle. Now you're just paying for audio and slightly longer run time with a real video camera.

eattheblinds

hi andy,
i'm the one in question and yeah, through my own trial and error / tweaking the settings, i found the one setting that made the most visible difference (i.e.. the one that eliminated really bad moire and aliasing) was the VBR setting dialed to its highest quality possible.  this was the one variable i made note of in my video since none of the other variables i tinkered with seemed to have the same significant difference.  i'm pretty new to both ML and the Raw hack, so if you wouldn't mind schooling me on what i've got wrong i'd be truly appreciative.
cheers,
paul

PS (if it wasn't clear) re: VBR i'm referring to menu within the 50D's resolution settings


Quote from: Andy600 on September 16, 2013, 04:02:21 PM
I've noticed a couple of blog articles from professional shooters recently that seem to contain mis-information about ML raw video.

and in videos that accompany an article by Paul Kell (also retweeted by @autoexec.bin on the front page of Magic Lantern) http://www.eattheblinds.com/2013/09/once-you-go-raw.html he clearly states that QScale VBR -16 has an effect on raw video aliasing.

Although I think these quotes are totally incorrect, what would make a professional come to these conclusions? Are they simply mistaken or is there any substance to what they are suggesting?

1%

QuoteVBR i'm referring to menu within the 50D's resolution settings

This setting has no effect until recording starts.. at that its just the QP parameter to the H264 encoder chip.

Danne

Phillip Bloom is still pissed for ML "ruining" his 550d a few years ago  ;D

eattheblinds

Quote from: 1% on September 18, 2013, 08:02:04 PM
This setting has no effect until recording starts.. at that its just the QP parameter to the H264 encoder chip.

once recording starts is all i should really care about, no?  when i'm setting this up the info warns (and i'm paraphrasing since my 50D is in the US and i'm in vietnam) that this highest setting is not recommended since i need very fast write speeds to operate at these settings -- does that not mean the bit rate is being ramped up?

1%

The raw video is pre H264/live view. The encoder never runs. If you're recording H264 then yea, I guess set it and picture style.It will have a slight effect on aliasing there. Mostly though higher QP preserves detail in the MOV file.

eattheblinds

i do understand that, but on the 50D raw video menu where these settings live, you can tweak the resolution.  i agree with you that they should probably only relate to h264, but when i've played with these settings the raw video bit rate goes up significantly (hence the diminished moire/aliasiing).  a shame i don't have my 50D with me to screen grab what i'm talking about since i probably just sound like an idiot.  bear with me while it takes me a couple weeks to get back to the US.

as is, i've stepped up to shooting raw on the 5DM3 and once i'm confident i've got a rock solid setup i'm planning to use it to shoot b-roll and talking head interviews for a pilot we're filming mid-october.  for the sake of keeping things simple our A & B cams are C300s.  i would love to be in a position to use 5DM3s MLd to shoot raw as our main cameras, but i'm not 100% confident in my knowledge and experience with the raw hack yet.  no bias against DSLRs shooting raw (it's incredible, as i wrote in my blog) i'm just not ready to stake my rep on it.  give me a few more months of practice and i'll be there ;) 

anyways, i'm hoping i can find a little guidance here with a recurring problem i'm having:

i keep being saddled with a 4.29gb ceiling on the files.  i haven't actually timed out these shots, but they're (obviously) a little bit longer takes.  once i attempt to ingest them into raw2dng i get an error notice telling me "This ain't a lv-rec RAW file".  it was shot just like all the other raw clips, the only difference was that it was a longer take.  is it possible to salvage what was shot and (most importantly) how do i avoid this problem in the future?

Andy600

@eattheblinds - hey :)

I'm pretty sure that VBR can't affect raw recording in any way but I'm very curious because you say bitrate increases and I do see the difference in your videos. I've tried it myself and there is no discernible difference in bitrate or appearance at any VBR/CBR setting which makes sense because, as 1% says, it only affects h.264 and Liveview pre-recording BUT if you're getting this result with raw video files something must be happening somewhere in the pipeline. Might it be your NLE output encoder settings?

re: the raw error/spanned files. The 50D doesn't support ExFat so it needs to cap filesize by silently starting a new file when it reaches the ceiling. These files need to be joined before extracting the DNGs. Some ML raw apps do this automatically (on Windows but not sure about Mac). I think there is a thread or several threads on the forum that explain how to do it.

Please do let us know your findings when you return to the US :)
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

eattheblinds

the error wasn't on the 50D, it's on the 5DM3 and it's just one file, no pairs. i followed the correct procedure to format the cards (komputer bay 1000x 64gb) exFAT. any ideas?