4K@60Hz(FPS)

[Rayzin]

yes, 8bit yuv420 4k 60fps h265 file is working nice (I've been saying so already) with no visible problems (a screenshot wouldn't help proving that, since most problems with most players is only stuttering, not artefacts or something that you could see on a static screenshot)

True but it would help reveal whether the RK3288 problem with displaying 8bit video is a hardware issue or not.

Just curious but what are the file specs of the raw file you are trying to encode? (size, YUV, format etc.)

[kaefert]

I'm joining together various different videos of various resolutions and framerates - thats what I use avisynth for - to unify resolutions and framerates, and to get smooth blending (function dissolve) between them.

It varies from 20MP Canon EOS pictures (though as of yet I'm only using the 8bit jpg's and didn't take a look at high bitdepth processing in avisynth, maybe I'll take a look at vapoursynth for that) to low-resolution dvd-quality video clips.

What do you mean would help reveal if a problem with 8bit video playback is hardware or software related? What problem do you mean? The only problem I have observed is that video framerates not matching the display-connection refresh-rate will produce stuttering (at least for 25fps video played via a 60hz hdmi connection)

[Batfink32]

I'm a bit confused about what you guys are talking about. My TV which has only 8bit 4k HDMI ports does not even give me the choice of 4k@60hz in the RK3288 settings, I can only get 4k@30hz option even with an HDMI2.0 cable. Are you guys talking about this or how videos are encoded?

[kaefert]

From what I've read in the previous pages of this thread, it seems that the
RK3288 does only support hdmi connections with 4k@60fps when the TV will accept 10bit color depth.


And from my experience when you are lucky enough that your TV does support that you will get stutter if the framerate of your video file does not mach this 60fps of the hdmi connection.

[Batfink32]

From what I've read in the previous pages of this thread, it seems that the
RK3288 does only support hdmi connections with 4k@60fps when the TV will accept 10bit color depth.


And from my experience when you are lucky enough that your TV does support that you will get stutter if the framerate of your video file does not mach this 60fps of the hdmi connection.

Yeah, that seems right. I find also that with 4k@30hz I get bad screen tearing. I believe this is a problem with the refresh rate not matching up to the what the TV is expecting?

[kaefert]

though it would be quite interesting to know if that lack of 4k@60hz@8bit hdmi support is hardware or firmware related

[Gormar]

My TV which has only 8bit 4k HDMI ports does not even give me the choice of 4k@60hz in the RK3288 settings, I can only get 4k@30hz option even with an HDMI2.0 cable. Are you guys talking about this or how videos are encoded?

Sorry, but this thread was hijacked when we achieved consensus that to use 4K display at 50/60fps we need TV which accept 10bit colour values in that mode. Then it is a shame to use videos encoded with 8bit color depth which will be interpolated before sending through HDMI interface as 10bpc. But as it came out encode video at 10bit color depth using libx265 isn't trivial task.

[Batfink32]

Sorry, but this thread was hijacked when we achieved consensus that to use 4K display at 50/60fps we need TV which accept 10bit colour values in that mode. Then it is a shame to use videos encoded with 8bit color depth which will be interpolated before sending through HDMI interface as 10bpc. But as it came out encode video at 10bit color depth using libx265 isn't trivial task.

Cool, thanks.

[Rayzin]

Sorry, but this thread was hijacked when we achieved consensus that to use 4K display at 50/60fps we need TV which accept 10bit colour values in that mode. Then it is a shame to use videos encoded with 8bit color depth which will be interpolated before sending through HDMI interface as 10bpc. But as it came out encode video at 10bit color depth using libx265 isn't trivial task.

Consensus was not achieved especially since I introduced reasonable doubt using actual data. Now if I misunderstood something go right ahead and correct me but doing so without actual data leaves you with a weaker argument from the outset. But it doesn't mean you are wrong. If you know something I don't then why don't you help me understand and show me where you get your information from?

Then we have kaefert who explicitly states he has used 8bit 4k@60 without any issues. Or did you not see that?

[kaefert]

@Rayzin - I never stated that I have an HDMI connection of 4k@60fps@8bit --> 4k@60fps HDMI connection will only work when on 10bit (from what I have observed) what we where talking about with 8bit and 10bit was about video encoding, not HDMI connection.

So what I'm trying to achieve is to get my videos encoded with 10bit to actually make use of all that extra room in the color space.

@Batfink32 could you maybe share a screenshot of that "bad screen tearing" - I don't know what you mean by that (I did not try to configure my HDMI connection with 4k@30fps though).
update: never mind Batfink32, I've looked at the wikipedia article what screen tearing means http://en.wikipedia.org/wiki/Screen_tearing (that sucks! lucklily I don't have it)

[Gormar]

Then we have kaefert who explicitly states he has used 8bit 4k@60 without any issues. Or did you not see that?

Kaefert didn't proof that RK3288 can output through build-in HDMI interface picture at 4K resolution and 50/60Hz in 8 bit color depth. He is using Samsung UHD TV which support 4K 50/60Hz and 10 bit color depth.

On the other hand no one officially announced that RK3288 HDMI interface cannot output 4K at 50/60Hz and 8 bit color depth but reports aren't favouring other conclusions.

As far as I know only one person (jeorentie) might use TV which as per instruction manual could support 4K 50/60Hz mode only in 8 bit color depth and yuv4:4:4 but even this isn't 100% confirmed.

After some more reading I might have another theory why some TV claim to support HDMI 2.0 interface but cannot cooperate with RK3288 at 4K resolution and 50/60Hz refresh rate. Maybe these TVs doesn't support 4:2:0 chroma subsampling but RK3288 can only use that and not 4:4:4 (no chroma subsampling). In accordance with HDMI 2.0 specification this resolution at that frame refresh rate can only be achieved with 8bit color depth and yuv4:2:0 or yuv4:4:4 or RGB. At 10 bit depth and higher from these two chroma subsampling modes only 4:2:0 is supported. Panasonic TX-xxAX630E instruction manual claim that in 4K at 60Hz HDMI mode it use no chroma subsampling (4:4:4) therefor if RK3288 HDMI interface support purely 4:2:0 chroma subsampling then only 30fps is possible and not even in 10 bit color depth mode but in HDMI 1.0 mode.

[Rayzin]

@Rayzin - I never stated that I have an HDMI connection of 4k@60fps@8bit -->

Ok!

ugh let's hope that the mass confusion has ended now!

[Gormar]

I've got some mixed thoughts about RK3288 and 4K at 60fps. It looks like even experts aren't sure about HDMI output capabilites.


For example Openhour Cameleon FAQ states that RK3288 only support HDMI 2.0 mode without chroma subsampling (4:4:4) and there is no mention about required 10bit color depth. But in that case Panasonic TX-xxAX630E should work with RK3288.
I think it is neccesary to completly reveal which modes RK3288 HDMI interface can realy handle. If that will not happen then it would be quite hard to assert compatibility between this SoC and any UHD TV and more misleading information will surface.

If someone whose UHD TV can work with RK3288 and 4K resolution at 50 or 60 fps would send me registry file from HDMI interface I will be glad and maybe capable to determine in which mode it is realy working.

To do so you must have root access and execute this command:

[I]cat /sys/kernel/debug/rk3288-hdmi/hdmi > /mnt/sdcard/hdmi_regs.txt
dmesg|tail -65 > [/I][I]/mnt/sdcard/hdmi_dmesg.txt
[/I]

and attach output files (hdmi_reg.txt and hdmi_dmesg.txt) here with new post.
Note: Before executing second command (pressing Enter) suspend your device for a moment.

Other option is to run my program which will display some HDMI interface settings read from register values stored in /mnt/sdcard/hdmi_regs.txt file. Download and extract atached file put it in the /data/ directory and set execution rights to that file. The enter command: ./RK3288_HDMI_reg.elf
Grab the outputed information and share it here.

[kaefert]

Hey there Gormar!

Here sys_kernel_debug_rk3288-hdmi.txt you can find the contents of

/sys/kernel/debug/rk3288-hdmi/hdmi

on my Beelink R89

And here dmesg_hdmi_snippets.txt are parts of the dmesg output of my Beelink R89 that looked relevant to the HDMI connection for me.

And I've got a question for everbody who is using an RK3288 media player with 4k @ 60Hz --> what HDMI cable are you using, and where did you buy it? The cable that came with my Beelink R89 does work in general, but still gives me signal drops sometimes (up to 2 seconds of a black screen). I've tried a few others, but all of them where much worse (more of 50% of the time signal drops & black screen) then the one that came with the box.

[Rayzin]

For example Openhour Cameleon FAQ states that RK3288 only support HDMI 2.0 mode without chroma subsampling (4:4:4) and there is no mention about required 10bit color depth. But in that case Panasonic TX-xxAX630E should work with RK3288.

It is all very confusing isn't it? The wiki was one of the main reasons why I looked into this more closely. Displaying raw 4:4:4 at 4k@60 sounded too good to be true and the HEVC decoder profiles according to the datasheet of radxa states only 4:2:0 can be displayed and no more 4k@60. But who is right?

My post here is to bring some bad news.

I contacted Rockchip and asked if 8bit was supported 4k@60 and the answer was no, only 10bit. I am waiting for a reply of follow up questions and I'll post them here. It's really disappointing for those who only have 8bit 4k capable tvs. But after all this confusion, lack of transparency, misleading information etc. I am glad to have closure on this issue.