Jump to content

4K 8bit to 2K 10bit - Let's get to the bottom of this please!


Guest Ebrahim Saadawi
 Share

Recommended Posts

jUnfortunately there is no free lunch if you want to get rid of these types of artifacts.. gotta go higher bit depth and lower compression.

 

A lot of videographers had the delusion, not corrected (for obvious reasons) by companies like Atomos, that 'clean HDMI out' would make their DSLRs cinema cameras. Although HDMI out is always 'uncompressed', it stays 8-bit 4:2:0 if that's what the camera records, even if the external recorder writes 10-bit as ProRes or DNxHD.

 

Everytime one is confronted with a shortcoming, he feels the urge to overcome it. With the issue of banding in the skies (and similar 'motifs'), workarounds are discussed.

 

We've seen a lot of okay-looking skies from DSLRs. But with MagicLanterns raw and split-screen comparisons, even on vimeo, one could clearly see that a banding-free sky is not the sky.

 

Avoid the sky? Live with some minor banding (saw it in quite some TV docs that were otherwise nicely photographed)? Downscale and dither 4k? Experiment and find a proper picture profile and proper exposure? Grade conservatively? Shoot raw? Genuine ProRes?

 

We are getting more demanding and critical and tend to see weaknesses clearer than strengths. The receipt to be unhappy. But of course our videos are getting technically better, the industry says thanks.

Link to comment
Share on other sites

EOSHD Pro Color 5 for Sony cameras EOSHD Z LOG for Nikon CamerasEOSHD C-LOG and Film Profiles for All Canon DSLRs

cpc- I've sharpened the 4K GH4 footage in post and the noise grain is very fine. There are smeared areas and macroblock artifacts in some places, though overall when there's not a lot of motion the noise grain is pretty impressively small, especially compared to my 5D3 (RAW) or FS700 (AVCHD).

 

Premiere uses a form of Lanczos and Bicubic for scaling ( http://blogs.adobe.com/premierepro/2010/10/scaling-in-premiere-pro-cs5.html ). Whatever they are doing appears to work reasonably well. In terms of the 10-bit luma debate, if the 8-bit 4K footage was effectively dithered, either via error diffusion or simply noise, then resampling a 4x4 grid  (Bicubic) could kind of reverse the error diffusion and put some bits back into luma. Intuitively it doesn't seem like it would buy a lot of latitude vs. native 10-bit sampling, however a little bit can be helpful. Adding error diffusion / noise certainly helps reduce the appearance of banding/blocking. Ideally the dither/noise would only be added where it's needed. Without significant dithering of some form or another, I don't see how 4K to 2K could do much for the '10-bit luma' argument as we need variance for the 4 source pixels to spread around the values of the summed/averaged final pixels.

 

But there is usually some variance, moreso with multiple samples per output pixel. Yes, noise provides this, but so does detail and even gradients can have it (the steeper, the better), especially without compression on top.

The thing is, in blacks and dark grays, where precision is most lacking, noise is at its strongest. So in practice you gain the most exactly where you need the gain.

(On a remotely related note: Been playing with some Kinefinity footage lately; noise is nice and the image scales down to 2K beautifully.)

 

@Axel: Not necessarily the case with HDMI out being subsampled the same as the in-camera recorded video. Lots of cameras record 4:2:0, but output 4:2:2 on HDMI. And some go 4:4:4 (over SDI), F3 comes to mind.

Link to comment
Share on other sites

I think that part of the problem is that when we think about "8-bit 4:2:0 1080p" we have a particular picture in mind, a perfectly reasonable one - a big rectangle of 8-bit numbers, each essentially independent. That is how our 8-bit monitor works - one number can be 256, the next one 0, the next 131, and so on. On that picture, banding will appear due to quantization when you try to represent smoothly varying material directly. (It will appear in 10 bit (or 12 bit, or 14 bit..) but the bands will just be smaller and more similar, so at some point it is imperceptible.) Dithering can decrease banding and increase perceived color range because our eye essentially averages over neighboring pixels, and throwing in a (semi)random 16 among a bunch of 15s makes us think we are seeing, say, 15.25.

 

The point Andrew keeps making is that what we get (out of non-raw cameras) is COMPRESSED 8bit 4:2:0, for example. And the way compression works is that it does not allow you to just randomly place numbers in boxes in the array. It removes "extraneous" detail to reduce information. So it would see that 16 among all the 15s and say "hmm - outlier - toss it." In the end, the quality of the implementation of the compression is going to be the big difference maker (as Andrew, again, always says), and so the 8-bit spec will be less important than the algorithms implemented.

 

If the final output is compressed in any of the standard ways, the dither will go away and the banding will return.

Link to comment
Share on other sites

@Axel: Not necessarily the case with HDMI out being subsampled the same as the in-camera recorded video. Lots of cameras record 4:2:0, but output 4:2:2 on HDMI. And some go 4:4:4 (over SDI), F3 comes to mind.

 

I received a newsletter from Canon, regarding the popular 5D M3 (my awkward translation from german):

 

 

Due to the feedback of professionals in film and TV production the new firmware for the EOS 5D M3 will provide clean HDMI out to enhance post production possibilities and meet strict technical standards. Through this interface high definition video can be transmitted in YCbCr 4:2:2, 8-bit.

 

4:2:0 ist not responsible for banding ...

 

EDIT: According to a friend, this firmware was already released, but there were doubts if it really was 4:2:2.

 

The video engineers once decided to compress to 4:2:0, because human vision can't tell the difference anyway.

Link to comment
Share on other sites

So if your idea of good looking is film, and even the digital camera companies use that as a benchmark, people need to remember that film is scanned for a digital intermediate at 10bit log encoded gamma to preserve the density information of the film negative, and so that it can be reprinted w/o any loss. 

 

Though just a couple of years (3-5?) ago, commercials I edited (that were shot on 35mm film) were transferred to us on Digibeta. That's 8bit 4:2:2 720x576 here in Pal land.

Link to comment
Share on other sites

 

Though just a couple of years (3-5?) ago, commercials I edited (that were shot on 35mm film) were transferred to us on Digibeta. That's 8bit 4:2:2 720x576 here in Pal land.

 

 

You're referring to film to SD broadcast, I was referring to a film DI, back to film pipeline, but I'm working on a TV show right now that came off digi-beta at 10bit 4:2:2 1080 24p HD, a D5 deck and it's pretty old (Panasonic introduced them in 94!). Digital beta can be either 8 or 10 and there are several varieties, including 10bit SD.

 

If you want to see an example of some 2k 10bit 4:2:2 digi-beta that came from an Arri Alexa, I used an old clip for background on a test in the Grading sticky, page 3.

Link to comment
Share on other sites

  • 1 year later...

I found this topic after searching info about banding and dithering. I have got very good results with my GH4 by using dithering filter when playing GH4 4k videos. GH4 makes quite much banding in skies but when adding dither in playback the dither pattern is perfect because the is no re-compression when playing files. GH4 8bit 4k videos look almost like true 10bit videos because the dithering filter "rasterize" more tones for eye than a 8 bit file can record. It is just like a printing raster which is 1 bit but can still show perfect gradients.

Link to comment
Share on other sites

I use Potplayer in PC. It has "deband" filter in video processing section. It has two settings: treshold and radius. It makes clever dithering to smooth areas and bands. I made an example frame with heavy sharpening so that the effect is more visible: 

http://***URL removed***/forums/post/57489279

Here is a another sample frame. Look at smooth gradients in water:

http://***URL removed***/forums/post/57494349

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

 Share

  • EOSHD Pro Color 5 for All Sony cameras
    EOSHD C-LOG and Film Profiles for All Canon DSLRs
    EOSHD Dynamic Range Enhancer for H.264/H.265
×
×
  • Create New...