|
|||||||||
|
Thread Tools | Search this Thread |
January 19th, 2009, 02:11 PM | #16 |
Regular Crew
Join Date: May 2005
Posts: 41
|
Where do i download and how do i get them on my camera?
|
January 19th, 2009, 03:15 PM | #17 |
Major Player
Join Date: Sep 2006
Location: Garden Grove CA
Posts: 239
|
Where it says flat +10 , flat +20 , and high gamma 5.0 under the pictures.
|
January 19th, 2009, 03:17 PM | #18 |
Regular Crew
Join Date: May 2005
Posts: 41
|
it just give me a .txt file. What do i do with that?
|
January 21st, 2009, 07:18 PM | #19 |
Major Player
Join Date: Nov 2003
Location: Chicago, IL
Posts: 439
|
it's not just the blacks... the highlights look clipped too...
here's the other link as well: http://prolost.blogspot.com/2009/01/...hing-news.html Last edited by Josh Brusin; January 21st, 2009 at 07:22 PM. Reason: add the link |
February 3rd, 2009, 11:28 PM | #20 |
Major Player
Join Date: Dec 2004
Location: New York City
Posts: 613
|
Canon Picture Styles
I'm surprised no one has started a thread on this yet. I mean the EX1 section has a big picture profile thread and it isnt even as tweakable as the 5Dmk2. The canon picture style editor gives you full control over your the images that come out of the camera. I have begun the process of working on several custom picture styles and have intentions of making multiple matching profiles for the 5Dmk2. I have a ColorChecker chart that I've started using, and am investigating the effectiveness of creating matching profiles for other cinema cameras such as the Panavision Genesis, Panasonic Varicam, etc. (and maybe some film stocks) based on charts which are available on websites like cinematography.net . Obviously this is not the ideal way to match cameras, but I am looking more to see what kinds of looks I can get out of the 5Dmk2 that are inspired by much greater cameras than to shoot side by side with an F23 and have matching footage.
I am, however, interested in creating accurate matching profiles with the Sony EX1 and Panasonic HVX200 since I own an EX1. I did a quick match with my EX1 using a crappy chart just to see the process, and I also made some style used to push or pull the exposure 1 stop ( which can be very useful for gaining extra manual control). With the push 1 stop style, you could use iso200 and expose it like iso400 (with highlight tone priority), which allows for shutter control even in iso400 lighting setups. I dont have any picture styles I'm confident enough in to post yet, but I will soon and here are some extras that are available from canon as well as instructions for creating your own: Canon : Picture Style One snag I have hit is that video from the 5dmk2 appear to have a gamma shift compared to photos. If you take a photo and a video with the same exposure settings, there appears to be about a ~1.2 gamma shift, making the video a bit brighter than the photo. This can be corrected in post, but its not clear if it is exactly a 1.2 gamma or if there is more to it. This is using Quicktime 7.6 and sRGB and I have tried the qtgamastripper utility since I know this is often a problem with qt h264 video, but it did not make a difference. This makes things a bit more difficult since building picture styles is done on RAW photo frames so they are not as useful if the video is not going to match. Anyone got any ideas, maybe im missing something? |
February 4th, 2009, 01:54 AM | #21 |
Inner Circle
Join Date: May 2006
Location: Camas, WA, USA
Posts: 5,513
|
The Picture Styles are cool. We used a custom style on the short film we just finished to good effect.
It doesn't really give "full control" though. For instance, before the "re wrap" fix was discovered, we tried to change the levels to 16-235 in camera with a preset. Long story short, the picture editor doesn't really give that level of control. I like your idea of using the styles to match other cameras though. This could be really helpful as many will want to use their "normal" cams for much of a production, and then break out the 5D Mark II for those low light and shallow DOF shots. Regarding the gamma shift, you are right that the latest Quicktime decoder messes with the gamma. We use the re-wrap trick and Vegas, then create Cineform proxies for editing. The final render is done (time allowing) from the re-wrapped originals. By using re-wrapped h.264, we don't get a gamma shift. Looking at the histograms, there are no gaps in the levels. It's smooth from 0 to 255. The QT output, however, does have gaps and bumps, and I don't believe that it's 100% recoverable, since the output is already truncated to 8-bits. With the effort that the camera matching picture styles would take, I'd strongly recommend using re-wrapped files. Not only does it give the best quality (255 unique levels, rather than 219), but it also removes an unknown variable from the process. Best of luck with your efforts!
__________________
Jon Fairhurst |
February 4th, 2009, 05:02 AM | #22 |
Regular Crew
Join Date: Jan 2008
Location: Johannesburg, South Africa
Posts: 35
|
Noah... What u said about pushing 1-stop in the pic profile is interesting cos like u said u can shoot at the equivalent of iso400 (with highlight tone priority) and gain the shutter control we all need! (this was a deal breaker for me as you all know Ill never shoot with that horrible slow-shutter look)
But the question remains - is it an artificial 1-stop push? - being equivalent to pushing it in post? (in that case we could just under-expose and push in post for shutter control = horrible grain/noise) OR... is it actually using the RAW data to adjust iso BEFORE the compression???... In this case I think you may have CRACKED IT!!! :) :) Now in this case the 5Dmk2 could be viable as a professional video tool! Someone needs to check if the pre- adjustment/brightness/iso in the color profile setting is PRE... OR post the compression. This test can be done by pushing the profile 4-stops in the pic profile... and then in-camera with the ACTUALL iso adjustment live!!! Pretty please will someone test it??? |
February 4th, 2009, 07:39 AM | #23 |
Major Player
Join Date: Dec 2004
Location: New York City
Posts: 613
|
The push 1 stop picture style certainly works in theory, and personally i would consider building all picture styles as pushed 1 stop because it potentially allows an extra stop of latitude in highlights, sacrificing at least 1 stop of shadow detail. The quick test I did with my push1stop picture style seemed promising but i need to do this more carefully because i got a little bit of a color shift and should crush the black levels more since noise becomes more of an issue with such a style.
It seems like a good idea to me since iso400 is more reasonable for shooting indoors. You'd choose this profile and set your exposure compensation to -1 to get correct exposure. It is not as good as hardware iso gain, but it is much better than pushing in post since the camera applies the picture style to the 14bit RAW images before compressing it to video. This software gain is similar to how RED handles ISO changes. Also, this is what the 5dmark2 does for iso values past 3200 (they are fake gain values past there based on a few things I have read). I've tried it with a 2stop push as well for is800 control but you really have to be careful about building the picture style just right for it to look right, I think it can be done though. The toughest part I've found is getting color saturation to match within these styles. I have heard this re-wrapping technique mentioned around but have not paid close attention since i thought it was just a way of getting around the studio RGB issues that i thought quicktime had ultimately fixed in 7.6 . When you re-wrap your videos, do you get RGB values exactly equal to RGB of jpegs shot with the same picture style and exposure settings? I dont think it would be right to build picture styles from video that is not being correctly handled, I will assume the video contains the same information as jpeg but that it just must be decoded correctly. |
February 4th, 2009, 07:44 AM | #24 |
Obstreperous Rex
|
Actually yes we've had some discussions about this already -- threads now merged.
|
February 4th, 2009, 07:45 AM | #25 |
Major Player
Join Date: Dec 2004
Location: New York City
Posts: 613
|
BTW when i say like RED, i mean software gain. Of course RED does not bake this in, whereas we have to because we are only given 8bit output.
|
February 4th, 2009, 10:34 AM | #26 | |
Inner Circle
Join Date: May 2006
Location: Camas, WA, USA
Posts: 5,513
|
Quote:
I know for sure that the re-wrap doesn't have the gamma lift that we see in QT 7.6. It presents all levels from 0-255 cleanly. I can't confirm that it's identical to the stills, since my son is a few hundred miles away with the camera. In fact, my assumption is that it's close but not perfectly identical. This is based on my shutter tests. For instance 1/40 and 1/50 are actually something like 1/45. 1/60, 1/80 and 1/100 are actually 1/80, and the gain varies between them. (1/60 is actually slightly darker than 1/80, and 1/100 is much darker.) Canon 5D Mark II Shutter Exposed on Vimeo (Note that this video was made before I was aware of the re-wrap. The shutter analysis is valid, but it suffers from the original black crush problem.) In any case, the 8 bit values are not corrupted with re-wrapping, but they are with the nominal QT 7.6 output. I think the re-wrap gives the best possible 8-bit output available from the camera. The difference between photos and video are a separate issue. You have no choice but to design the profiles with RAW photos, but you will need to confirm the results with still captures from video. And it's possible that you'll need to make a different set of profiles for each shutter speed and over a few ISO ranges.
__________________
Jon Fairhurst |
|
February 4th, 2009, 02:28 PM | #27 |
Regular Crew
Join Date: Jan 2008
Location: Johannesburg, South Africa
Posts: 35
|
Nice explanation Noah, just to clarify you saying its just a good 14/bit Raw 1-stop push and not the "REAL" iso adjustment? Even so is the noise noticable?... how about 2-stop push?
|
February 4th, 2009, 03:01 PM | #28 |
Major Player
Join Date: Dec 2004
Location: New York City
Posts: 613
|
Right, its just reassigning values in software, not hardware gain. This means it will lose some dynamic range (depending on implementation) to be completely accurate with highlights clipping and all, youd end up losing up to a stop of dynamic range in the shadows. But that would be throwing away some highlight data, which could just be compressed and retained although then it wont completely match things shot in another style. It is just another option, if you want to shoot in real hardware iso400 you're stuck with a 1/40th-1/50th shutter (if that is really what it is using).
I was reading somewhere that some of the 1/3rd stop iso intervals might be achieved in software gain and only full stops or certain 1/3rds are achieved in hardware, but I have yet to verify this. Well heres a picture style i was working on based on the Panavision Genesis Panalog4 chart tests on cinematography.net . I haven't had a chance to play with it that much and it is sort of work in progress. I think It gets a little too green in the highlights or something, but im not sure. Genesis Panalog4 Picture Style (Work in Progress) http://noahyv.com/temp/picturestyles...nalog4_WIP.pf2 I used DPXs from this test posted on cinematography.net : Cinematography - CML is a place for the discussion of Professional Cinematography along with shots of my own colorchecker to work on this. Obviously the conditions were very different but I wanted to see what I could get from it. Lemme know what you think and if there seems to be anything I am doing wrong or can do better. I want to make sure I get some practice in this before I trust myself to do a good matching style with my EX1. I started doing one based on some frames from some tests of Film stocks but there was not enough information (not enough under-over exposure steps) to get enough data, so was forced to try another method and the curve came out wrong. |
February 4th, 2009, 03:08 PM | #29 |
Major Player
Join Date: Dec 2004
Location: New York City
Posts: 613
|
BTW regarding noise, I have found banding noise to be a bit noticeable in video, definitely more so than in stills... perhaps it is from sensor heat? It wouldnt be surprising at iso3200+ but i think i saw some banding noise in underexposed video shot at isos as low as iso200... I'll try to do some tests to verify. Obviously, pushing or pulling footage would only make this worse. From just a little use, i would say a 1stop push should be pretty usable as long as shadows are kept in control to crush noise a bit. 2stops might be too much but I think with this amazing sensor and a little finesse it could be quite nice. Of course 2stop probably wouldnt be able to look clean and match footage shot in other styles particularly because of some limitations of canon's picture style editor.
|
July 1st, 2009, 07:14 PM | #30 |
Major Player
Join Date: Dec 2003
Location: Fort Worth, TX
Posts: 414
|
What is this "re wrap" you speak of?
|
| ||||||
|
|