jcs

jasonmillard81- the models are indeed beautiful; the GH4 with the Natural profile does a nice job with skintones. The 1.x firmware on the GH4 was a little too overprocessed/clean and required a bit of post work to look more filmic and less like video. Haven't tried the 2.0 firmware; some reports are that it looks more filmic/less-processed straight from the camera. All the GH4 shots were 1080p 60fps VFR mode which is probably closer to 720p in terms of actual resolution. Here's a 1080p vimeo version (just one model):

The GH4 and the Natural picture profile provide decent skintones: https://www.youtube.com/watch?v=Hc2LXH_MsQQ https://www.youtube.com/watch?v=Hc2LXH_MsQQ (Shot on the GH4, GoPro Hero 3, and iPhone 5S) We recently purchased FilmConvert: it has a GH4+Natural calibration and it does a pretty good job moving the GH4 into a more filmic look. We prefer the color of the A7S with a tweaked Slog2 or CINE picture profile for 1080p (and 5D3 RAW is still the best for skintones), however the GH4 is pretty nice for 4K with a little post work. It will be interesting to see Nikon's entry to 4K video.

This is nice bar for for other cameras to match or exceed shot on 5D3 RAW:

Hey John- I submitted a detailed bug report as well. Since the OS GUI locks up too, it appears to be both Adobe and NVidia issues (the GPUFoundation.dll crashes are Adobe, however could be caused by NVidia drivers. That said, Adobe would need to work around driver issues). My temporary solution? Boot my 2010 MacPro normally running Windows 7x64sp1 into OSX 10.9.5. No more lockups or crashes so far (but occasionally black screen (audio still works)). This happens less often, so I can work faster than in Win7. When there are no issues, Win7 runs faster than OSX (same hardware). I upgraded a Quadro 5000 to a GTX 770 modded for OSX by macvidcards: major boost in editing and rendering speed. When the GTX 980 is more readily available, I'll upgrade to a macvidcards modded GTX 980 (165W for the 980 vs 230W for the 770, and of course faster too). I wouldn't expect a fix from Adobe until late October when CC 2014.1 is due to be released (perhaps pressure from users and Adobe could get NVidia to fix their end sooner- they release drivers frequently). Didn't get any replies from Adobe folks on their forum for this issue.

The lockups/crashes don't happen in OSX (10.9.5)- same hardware (doesn't run quite a fast overall (CUDA is much faster than OpenCL), but is a much better editing experience without the crashes and lockups).

Premiere Pro CC 2014.0.1 with NVidia (CUDA) on Win7x64 locks up / crashes (GPUFoundation.dll) with 4K material (green screen). Detailed info here: https://forums.adobe.com/thread/1586611 Anyone else experienced this- any solutions?

It's unfortunate that Canon isn't upgrading their DSLR video quality to be competitive. Lots of us have many Canon lenses and being able to shoot very high quality stills as well as video with a single camera is very, very useful. If the 5D3 successor had an XAVC 422 10-bit class codec with log encoding and at least 5D3 RAW-level resolution, it would rock. Between the A7S and GH4, 5D3 RAW produces superior color and video quality (except extreme low light and of course no 4K). Good for Nikon and the D810- folks with Nikon lenses can now have an excellent stills camera with autofocus that shoots nice video. Regarding the post down sampling-low pass filter comment to remove moire/aliasing: if that were possible we'd have plugins which remove aliasing in our NLEs. Mathematically, it's necessary to perform low-pass filtering before downsampling to prevent aliasing during downsampling. The only way I'm aware to remove aliasing/moire in post effectively is via super-resolution algorithms, which haven't been commercially available (research stage or only in limited hardware/embedded systems).

We can visualize what the SGamut to sRGB matrix is doing by drawing an untransformed basis (unit matrix) and also drawing the SGamut matrix (shown in brighter colors- a quick example in 3DSMax): A pure red input pixel is the 3D vector {1,0,0}, and we if we perform the dot product for each row we see the new coordinates (color): 1.87785, -0.17681, -0.02621, which matches the 3D drawing. The 3 points (darker colors) which represent the unit matrix (no transformation) also represent the edges of the color cube for a 3D LUT (which contain more points making up a full 3D cubic lattice). If we transform the points which make up a 3D LUT using the matrix, the resulting transformed cube will now exactly replicate (within float precision) the 3x3 matrix. Similar examples: By increasing the density of the 3D cubic lattice, we can also accurately perform a gamma curve at the same time as the matrix transform. The accuracy can be high when the 3D-LUT uses tricubic interpolation. Premiere Pro doesn't appear to have a GPU accelerated method for matrix operations (Channel Mixer), so a 3D LUT which combines the matrix with the the gamma curve will not only be much faster (GPU accelerated), but also easier to use (just one effect). It will also be plenty accurate (you shouldn't be able to see a difference, even with pixel peeping). Putting the final result in a .cube file will allow the transformation to be used in any modern NLE/Color-correction tool. dhessel's LUT is already a .cube file (representing a 1-D transform): LUT_1D_SIZE 1024 -0.00901794 -0.00901794 -0.00901794 -0.0088501 -0.0088501 -0.0088501 -0.00868225 -0.00868225 -0.00868225 -0.0085144 -0.0085144 -0.0085144 ...

Hey sunyata- a 3x3 matrix only has rows and columns which sum to 1.0 when it's a pure rotation matrix (no scale or shear) AKA orthoganol (orthonormal). A 3D LUT can perform precisely the same function as any 3x3 matrix. Visually, the 3D cube lattice points are each transformed by the matrix. When points (colors) are looked up in the transformed cube, the results are the same as the direct matrix computation (within float round off error). Now, if we transform the cube by a gamma function, we've got both steps baked into one transform, which can then be easily applied using the Lumetri effect in PPro. It's also very fast- GPU accelerated.

When performed on a GPU, LUT interpolation via a cubic spline (Catmull-Rom, etc.) in floating point will have more than sufficient accuracy vs. a direct lookup table with no interpolation. In other words, a 4-point LUT with spline interpolation can look the same a 256 entry direct LUT for 8-bit (simple gamma curve). Before GPUs, the direct lookup table on the CPU was the better choice for performance reasons. Current GPUs can perform tricubic interpolation on 3D LUTs extremely quickly, so the performance and ease of use (saving time) is worth the minor differences vs. a 1-D LUT + affine transformation. A 64 3D-LUT is 2^6^3 = 262K entries which is overkill for most uses, especially if only needing a gamma curve + affine transform. All that's needed is enough points to represent the spline control points for the gamma curve- perhaps an 8 3D-LUT (2^3^3 = 512 entries).

The example images look pretty good! One of the challenges with S-gamut is the non-linear response (color changes with exposure), right? S-Log3/S-Gamut3 fix this issue (doesn't change color with exposure). Since you're using the Lumetri effect to load a LUT (which will be fast and hardware accelerated), why not combine the affine color transform with the 1-D LUT into a 3D-LUT? Perhaps possible with SpeedGrade or Resolve? What's the advantage of using S-Gamut vs. Pro or Cinema color mode? (I'm currently using Pro color mode with S-Log2).

Of those three videos, the D810 example show the best color, resolution, and skintones. The 1DX looks soft (similar to Mark III) and color is OK, and the A7S color isn't pleasing (however this could be personal preference of the editor) is a bit noisy. Hard to compare when each video is very different- lighting, time of day, subject, different editors etc. Hadn't looked at the D810 much before: it's pretty nice (I also find auto-ISO useful (available on A7S too).

The Metabones Smart adapters v3 and v4 lock up the a7s on startup, typically after a lens change. Perhaps related to some kind of internal memory on the adapter. The workaround seems to be unlocking and twisting the adapter, powering up the a7s, waiting for the display to become active, then twisting to lock the adapter to the camera. Others have reported this issue and sending back to Hong Kong at $35 each time for a firmware update hasn't solved the issue. So far it's happened with our 24-105 and 16-35 Canon L lenses (not a third party lens issue). http://www.dvxuser.com/V6/showthread.php?325712-Problems-with-Metabones-EF-to-NEX-Adapte-Lens-doesn-t-work-until-battery-is-removed/page2 I haven't seen this issue with the gen I SpeedBooster yet.

We have a GH4 and the aliasing in that video looks like 1080p for this camera. We pretty much only shoot in 4K and have never seen aliasing like that. Looking at the test charts shot by Adam Wilt: http://www.dvinfo.net/article/acquisition/review-panasonic-dmc-gh4-camera.html/2 also shows only minor high frequency aliasing with 4K scaled to HD, whereas 1080p has aliasing (quote from Adam Wilt's article for the GH4 1080p chart):

hmcindie- the skintones are right per the vectorscope- it looks right on all my monitors, MBP, iPhone, etc. The best way to improve the color on location would have been to illuminate the skin with a broad spectrum light. What exactly is wrong with the color? How would you improve the color for this shot in post?

GMaximus- it would be cool if cameras had built-in software which could process a color calibration chart: shoot chart (pattern recognition lines up all the cells) then get accurate colors for those exact conditions. Pierre_move- my favorite film stock is Eastman EXR 100T 5248 (Men In Black II, The Last Samurai, Fight Club, Kill Bill, Lost, Rear Window, more info here: http://shotonwhat.com/film-negative-stock/eastman-exr-100t-52487248?view=more). Would be nice to get that look straight from the camera. Regarding the A7S still- I'm guessing the light source was sodium/mercury, etc., with a limited spectrum. The colors (lights) in the background are accurate. Here's another version where the skintones are precisely aligned to the skintone line on the vectorscope: How would you change the skintones for your preferences?

lafilm- 4K downsampled to 1080p will effectively be low-pass filtered (by most modern NLEs) and have very little aliasing. While it's possible that its 4K scaled down to 1080p and way over-sharpened, the aliasing looks like the 1080p from the GH4. A highly aliased image will artificially appear extra sharp. Most high-frequency noise will get removed by the relatively low bitrate of H.264 online.

Given the aliasing, it wasn't shot in 4K. Regarding noise, perhaps it's not really that dark, ISO wasn't very high, and perhaps Neat Video or similar was used.

Canon cameras sell very well because they make people look great by creating very pleasing skintones. Apparently for technical reasons (silicon sensors, etc.), Canon makes skintones look great at the expense of color accuracy. ARRI is the undisputed* King of Color Science and reading about how to make a C300 somewhat match an Alexa may be helpful in setting up new cameras such as the GH4 and A7S to produce more pleasing skintones (at the expense of color accuracy): http://provideocoalition.com/aadams/story/color-matching-a-canon-c300-to-an-arri-alexa Art noted that digital cameras do the best in daylight (~5500K) due to sensor response to blue. While I too have noticed that Sony, for example, does much better in sunlight, I've found that cameras also do well in Tungsten (~3200K). While Tungsten doesn't have much blue, it is a continuous spectrum light, and that I believe is the most important issue. Cameras have the most trouble in mixed light (fluorescent + incandescent + LED, etc.) or light sources with large gaps in the spectrum (some fluorescent and many LED lights). I had no idea what kind of light this was, however AWB with the A7S wasn't too far off (minor WB adjust in post: still a bit magenta but that's the look I wanted): In the Art Adams article above, he had to make custom profiles (mostly the special Color Matrix) for each desired color temperature (two provided). This makes it clear that getting pleasing (and/or accurate) color for a wide variety of color temperatures and light sources is a very challenging technical problem. Face/skintone detection is one way manufacturers try to deal with the problem (regardless of race, all skintones have similar color properties (along the 'skintone' line on the vectorscope: ). It's not likely there's anything we can do to picture profiles on cameras to improve color performance vs. factory settings for general use. If we tweak colors for a particular WB and light source, it won't be correct for another light source with different spectral properties. Thus, if someone posts a tweaked color profile that changes WB colors or general color matrices, keep in mind those tweaks might not work correctly for your particular situation. * this is the internet, so the likelihood of disputing is much greater than zero. The Sony F35 and F65 can produce great color, and the new RED Dragon and associated color science is looking very good. The camera system most people prefer in terms of color is the ARRI Alexa (now +Amira).

While we have the A7S and 5D3, Canon's superior color science for skintones makes the 1DC a desirable tool for stills and 4K video for those shooting professionally.

Pretty cool- a little bit of smash up derby too.

Thanks Chris! Elena replied: "Aaawww thanks Chris! I think the best part is the space battle action in the back :) " For those who haven't seen it, >Sea Bass was shot on the GH4 in 4K, our A-Cam for green screen studio shots.

Not doing anything extreme in post- no banding issues yet. Using Slog2 with Cinema Color Mode, +8 Saturation, and Cine2 (now using Cine1 as it's full range) with Cinema Color Mode.

Andrew, while it's relatively easy to get great color and especially skintones from 5D3 RAW, the same cannot be said for the A7S. While the A7S can look great, especially with post work (sometimes a lot of post work), it has a problem with green that can make skintones a challenge. Have also seen examples online where the highlights make skin look like plastic (something I don't recall seeing with 5D3 RAW examples online (certainly not with anything we have shot)). When the A7S can produce something like this:

Again, the luma curve (gamma) doesn't appear to be the issue with color- it's the Color Mode: S-gamut, which appears to be coupled to WB (from other docs and forum comments). That means it appears S-gamut changes with WB (the Slog2 version, not the Slog3 version which does not change with WB). Kholi on dvxuser swapped out S-gamut and used the Pro Color Mode and folks are getting good results without the yellow/green issue (he made other changes as well). I also found Pro to work well however to our eyes we like Cinema Color Mode better. While we can tweak things in camera and in post, it would be helpful if Sony could release full specs (whitepaper) for the A7S's Slog2+S-gamut along with any LUTs for use with common NLEs (PPro, FCPX, Avid, Resolve, etc.). The limited docs state that the A7S's Slog2+S-gamut (8-bit) is not the same as the pro version (10-bit).