Personal View site logo
Make sure to join PV on Telegram or Facebook! Perfect to keep up with community on your smartphone.
HELP - Getting CINESTYLE results from the GH2 ?
  • Well, I'm stumped.

    As a new & and proud owner of a GH2, I've downloaded the latest (V1.1 25p) firmware and am investigating hacks.

    A higher bitrate is obviously desirable - around 50Mbps will hit the spot, so I'm currently trying: "Driftwood Quantum X v5b and v5c 'β Pictoris' (baby Quantum X) VBR" (35 - 75Mbps variable)

    The REAL question is about capturing more dynamic range and squeezing maximum latitude out of the GH2 to get results that resemble Cinestyle. I see evidence here...

    ... yet trying that hack, I see zero latitude difference.

    Come to that, I've tried several hacks now, and they all look identical (apart from bitrate).

    I'm trying to research this, but struggle to get real answers. What should one do to get Cinestyle results as above???

    Thanks in advance...

  • 57 Replies sorted by
  • Some lighting ?

  • The hacks have no real input on the dynamic range. What picture profile are you using? What makes it look well is that this footage is properly exposed, within the dynamic range of the camera. Check out @shian's amazing tutorial on how to expose the GH2 well for grading here: http://personal-view.com/talks/discussion/3773/expose-in-the-zone-getting-the-best-image-for-grading./p1

  • I think you´ll get approximately what you want by using picture profile smooth or standard and dial down contrast, sharpening, saturation and noise reduction.. Some use -2 on all, but you have to try yourself to see what you want. Great link from Gabel, too.

  • Try some diffusion filters in front of the lens to knock down the contrast (and make sure you're flattening out your picture style to -2 or thereabouts).

  • Don't forget that movie productions use vast amounts of lights, bounce materials, light blocking materials, color gels, ND materials, time of day, etc, to get things to look good. Don't expect to get anything even close without at least using some of these.

    In the case of movies, dynamic range is totally overrated. Dynamic range isn't what makes things "pop", that's "contrast" that does that. contrast can be as small as lighting up a face to make it brighter than the body or background. In effect, you are actually DECREASING the dynamic range of the frame but introducing more contrast.

    However, in that clip, you can clearly see that the girl's face is illuminated by something from the left. Probably a bounce card or small light, or light from a building.

  • @Electric_Haggis

    The hack has no affect on dynamic range. You need to control the lighting or work some magic in post processing.

  • If you want it to look like Technicolor's "Cinestyle" then you'd expose properly and then raise the midtones with a levels adjustment in post. There's detail in the GH2 shadows (see video below).

    The real question is why you'd want it to look like Cinestyle. Cinestyle isn't supposed to be a look for delivery -- it's supposed to retain detail in shadow areas and highlights for work in post.

  • You can also use 5dtoRGB....

  • @qwerty123: Very true, you loose a lot of dynamic range from using AVCHD straight.

  • @Gabel

    false. false false false false false false false

  • @Fohdeesha I think one false suffices :-)

    @Gabel You don't loose dynamic range. The dynamic range is what it is (about 8-9 on the gh2). You could get the same luminance rendering (aka "cinestyle look") as 5dtoRGB with adjustments in post.

  • @onionbrain "The real question is why you'd want it to look like Cinestyle. Cinestyle isn't supposed to be a look for delivery -- it's supposed to retain detail in shadow areas and highlights for work in post."

    I disagree. I've used the cinestyle look on many short films. Punchy high-contrast images are quickly becoming passé. Even when I don't want the cinestyle look, I prefer beginning with cinestyle, as it's the easiest to get to look how I want it to, and the easiest to match with other footage. I never crush my blacks. Ever. I absolutely hate that aesthetic, and when a client wants it, I convince them otherwise.

  • @Sangye While I'm not a big fan of the high contrast, over-saturated look either did you use Cinestyle when shooting an already very flat scene? A friend uses Cinestyle exclusively on everything and some of his stuff looks quite dull even after post if he has shot a low contrast scene with it. It can be hard to pull reasonable blacks and whites without screwing up the mids if it is too flat.

  • Definitely! If you shoot a flat scene too flat in 8 bit there is no way pulling it back to more contrast without getting massive banding and macroblocking.

  • @qwerty123: Allow me to rephrase that, using a Quicktime based editor (Premiere Pro, Final Cut) with AVCHD does "lose" dynamic range, in that it can't handle all the info. So the dynamic range from the camera is the same.

    So my statement wasn't untrue, it was just worded badly, as Quicktime based players or editor won't be able to get the full luminance curve, hence the dynamic range is "smaller" unless you either convert or fix it (as info is lost otherwise).

  • I did this test here http://www.personal-view.com/talks/discussion/1795/dynamic-range-test-between-gh2-and-7d/p1

    My conclusion is that the hack does not give you better DR. But it preserve the most of what is already there. The other camera suffer because the low bitrate forces the codec to compress details that are less visible in priority like shadows. That is why the Cinestyle picture profile boost the shadows before they are being compressed by the codec engine.

    Another to note is that in raw the gh2 only trails the Canon Apsc and 5D by only half a stop of DR. Just look at sensorgen and dxomark. It is only the Nikon and Sony based sensor that have about 2.5 stop of DR.

  • @Gabel -" Allow me to rephrase that, using a Quicktime based editor (Premiere Pro, Final Cut) with AVCHD does "lose" dynamic range, in that it can't handle all the info. So the dynamic range from the camera is the same, but in post, some is lost unless you convert it or use this fix Andrew came up with..."

    Sorry, but with that claim, you're only digging your hole deeper. While Quicktime has had a history of gamma pitfalls, that did not cause any loss in dynamic range, only an incorrect interpretation of image data. Neither Premiere Pro nor Final Cut 7 are "based" on Quicktime; they use it only as a codec interface mechanism. To get a deeper understanding of these issues, I'd suggest delving further into the documented history of Quicktime's problems:

    http://provideocoalition.com/index.php/aadams/story/quicktime/

  • cinestyle hmm first try to have some reference of what you want to achieve, then research to see how the film or look was achieved, afterwards see with what resources available to you it is possible to achieve certain look, (camera moves, lighting, music, etc) cheers

  • Shooting during the "magic hour" or on an overcast day helps. It's natures diffuser.

  • @LPowell: Thanks for the link. I just tried to explain the problem as I have understood it. The cause I haven't been too keen about, I've cared more about the results. My point still stands: Using AVCHD straight for grading leads to loss of information.

  • @Gabel - "Using AVCHD straight for grading leads to loss of information."

    LOL, I'll leave you to your "loss of information".

  • @LPowell: I might be wrong, so please do correct me. I'm basing this on information Andrew (EOSHD) has found and comparing transcoded footage vs. raw AVCHD files in Premiere Pro, After Effects and played through Quicktime 7 (using Panasonic's AVCAM Importer). The difference was very big, with the transcoded ProRes files seeming more flat, having more detail and shadows, seeming to have at least a stop or more of information that was crushed/blown out on the AVCHD file. But as I seemed to have become a laughing stock based on my claims, so then do I ask, are these findings wrong? These are again unrelated to the hack and would be playback issue that I see on OS X.

  • @Gabel

    I haven't had time to investigate this myself. So I can't comment one way or the other.

    I haven't observed this, but I never work in 8bit, which these numbers clearly come from.

  • Yeah, I'm with @LPowell on this one.

    I can understand where you, @Gabel, as an intelligent and reasonable professional could be led in that direction, especially if there was some other context concerning the use of uncompressed formats with external recorders. And, the workflow you cited is, essentially, mediocrity that was archaic two years ago -- and could have yielded strange conclusions along the lines you mentioned due to a range of possible issues.

    But -- CS6 is a different universe from CS4 relative to AVC/mpeg-4 decoding/encoding -- and I can tell you (based on multiple conversations spanning several years with the Premiere Pro development team) that there's not a "loss of information" when you edit/correct/export AVCHD from Premiere Pro CS6.

    And, as you certainly already understand, the transcoding process certainly isn't inventing any information that wasn't already in the camera's AVCHD stream. It's the opposite -- transcoding can introduce various issues -- especially if the process wasn't managed perfectly by the user.

  • @Gabel Andrew had not done his research in that article, although he touches on an interesting subject. 5dtoRGB does a gamma lift.

    Premiere works in .709 - you don´t loose anything, except for worse output rendition (from h264) in versions prior to 5.5 (and it might be advisable to export through AE and / or do 5dtoRGB conversion for IQ critical projects). For instance you can notice interlace-like artifacts in high contrast areas on the red channel. You´ll notice how the detail is there, quickly, if you do a lift in premiere on a native h264; voilá enter shadow detail. There is also a feature in Premiere I recently learned about but have not explored fully; "Maximum Render Quality" which is set to off per default in the timeline. I usually select it in the export options but some tests would be appropriate for it.

    Output rendition is tricky and a bit erratic, directly from Premiere. The most reliable way to do it (if you do it the quick way) IMO is to export to prores and do a compressed file from that in media encoder or similar software.. But that is not "scientifically proven" to yield the best results in any way; it just has the least hitches in my experience.

    If you output a .709 file and display it in a full range player it will look washed out (compared to what you see in the editor); the blacks not really black..