Welcome to our community

Be a part of something great, join today!

  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

The limits of Redcode/Redray

Michel Hafner

Well-known member
Joined
May 21, 2007
Messages
376
Reaction score
0
Points
0
Hi,
I bought the Redray to encode my own computer generated 4K material. I hoped it would work as with 'normal' material from digital cameras or film scans.
It seems I was too optimistic. I realise Redray is amazing for what it is compared to the current state of h265 or other codecs. But it can't do wonders.
My material has even at 36 Mbit/s visible compression artifacts. I could live with that as the artifacts are not very strong and h265 at this rate is unwatchable.
What I can not live with is the fact that my material is not compatible with the real time decoding capabilities of Redray, it seems. The player can't keep up.
There is stuttering, blackouts, distorted sound and other anomalies coming and going with image complexity. I tried playback on a Dell 4K monitor and a
new JVC 'pseudo' 4K projector. On the monitor only the stuttering is visible. On the projector there are also the blackouts etc.
Does anyone else have similar experiences? Any solutions possible in addition to 'dumbing down' the source (e.g. remove detail).
 
The only blackouts I've had was solved by purchasing better quality "high speed" HDMI cables. This also effected the audio.

I haven't had any stuttering during playback however. That's a new one for me.

I'd be curious to hear more about the artifacts your experiencing at 36Mbps.
 
I'd be curious to hear more about the artifacts your experiencing at 36Mbps.
Clean smooth gradients next to high detail areas show compression noise. The high detail areas are 'chaotic' enough to mask any compression noise, The smooth gradients show issues without mercy.
 
Clean smooth gradients next to high detail areas show compression noise. The high detail areas are 'chaotic' enough to mask any compression noise, The smooth gradients show issues without mercy.

Just out of curiosity is your source material 16 bit?
 
Just out of curiosity is your source material 16 bit?
8 bit. But the noise is not there in the uncompressed version. Anyway, the issue is the stuttering. The noise I can live with,
 
All the preloaded clips and new examples show size 4096 * 2304. How exactly does Redray output to a UHD device? Scaling real time to 3840 * 2160? Cropping real time to 3840 * 2160? Leaving it to the device to scale/crop? Could the decoding plus reformatting exceed the processing limit? I think when I set central crop to UHD size for encoding my 4K files there are less problems during playback. Does that make sense?
 
Clean smooth gradients next to high detail areas show compression noise. The high detail areas are 'chaotic' enough to mask any compression noise, The smooth gradients show issues without mercy.

Any chance you can show some samples? Even provide an uncompressed frame or two that we can see if it's an issue to duplicate? I've been trying to break the compression and while it's not perfect, it's pretty darn close and I haven't run into any issues like you're describing.

Have you contacted RED?
 
I'll contact Red about this. The stuttering only happens when encoding with no crop. When cropping to UHD glitches can happen but are far more rare.
So does anyone know: How exactly does Redray output to a UHD device? Scaling real time to 3840 * 2160? Cropping real time to 3840 * 2160? Leaving it to the device to scale/crop?
??
 
I've been mastering everything at 4096x2160 at the moment and placing my 16x9 within that container, which provides the correct aspect ratio and content on UHD displays as far as I can tell. If you notice and export 3840x2160 you are losing some of the side content as it's "fitting to width" to the 4096 dimension. At least that's been my experience on most displays.

As for the source content. I would really need to see what the 8 bit stuff is. That is likely the culprit when it comes to gradient related details. Also I need to know more about the source imagery.

The best quality results I've been able to achieve are rather similar to how you would get the best results mastering a DCP. 16 bit source image sequences are useful as you are giving the encoder more of a "choice". Whereas 8 bit to REDRAY 12 bit might be a little less. That's my theory at least.
 
Update: Red has got test samples that show the problem. They see the problem too. They are looking into it. So far no feedback what's causing it.
 
Update: Red has got test samples that show the problem. They see the problem too. They are looking into it. So far no feedback what's causing it.


Curious to see what they find.

I am about to throw a ton of external 4K material into RREncode for future 4K Digital Signage.

I cant afford for this to fail.

I am still in the dark why my REDRAY is intermittently not playing all the clips on my system.
 
Hi,
I bought the Redray to encode my own computer generated 4K material. I hoped it would work as with 'normal' material from digital cameras or film scans.
It seems I was too optimistic. I realise Redray is amazing for what it is compared to the current state of h265 or other codecs. But it can't do wonders.
My material has even at 36 Mbit/s visible compression artifacts. I could live with that as the artifacts are not very strong and h265 at this rate is unwatchable.
What I can not live with is the fact that my material is not compatible with the real time decoding capabilities of Redray, it seems. The player can't keep up.
There is stuttering, blackouts, distorted sound and other anomalies coming and going with image complexity. I tried playback on a Dell 4K monitor and a
new JVC 'pseudo' 4K projector. On the monitor only the stuttering is visible. On the projector there are also the blackouts etc.
Does anyone else have similar experiences? Any solutions possible in addition to 'dumbing down' the source (e.g. remove detail).

So it isnt perfect, as people were argued at 10mb/s. I argued it was likely they would have to resort to these data rates to achieve this, so I am surprised there are still visible problems. It should be easy enough to reduce these artifacts out, leaving less of the image, but with no non image artifacts. This is half as good as I expected. How does it compare to top quality h265 (pity there is no ambarella encoder yet)?

The player itself, as far as stuttering etc, might just be the one in a thousand or ten thousand that is not like the rest. Maybe there might be some sort of frame feed back mechanism in HDMI requiring a better cable, but you should get freeze frame and dropped frames instead of stuttering (freeze frame without dropped frames).
 
There are indeed glitches with the REDRAY player. Video quality is pretty darn good, though. Honestly you have to go for a 4K DCP if you want better. As always, there are a lot of pixel-peepers and complainers, but I'll be the first to admit that I occasionally see compression artifacts or playback oddities. The product just wasn't finished. It has compatibility issues with various TVs and AV receivers. REDRAY does not play nice with the Denon 4K receivers, including the new X3100W and X2100W units that only shipped in the past few weeks. To pipe REDRAY through to my projector and to the Samsung 4K display in my office, I have to run it through a 4K matrix switcher and then use the separate audio out into the receiver. Shouldn't even need the matrix involved at all, but the video drops in and out when going through the receiver. Sucks because the receiver runs my office as a second zone, I don't even want to own that matrix switch at this point.....

There are apparently no updates coming for the REDRAY and it is EOL. Sure, we can still buy them, but it either works for us as is, or it doesn't. I really don't think RED should quit on this one. Lots of potential here still, although without HDMI 2.0, HDCP 2.2 support it's a non-starter at this point. And it will have to offer some reasonably competitive media library at a price point similar to Sony. Otherwise no one is going to buy. I can encode H.265 4K content and play it back on the Sony FMP-X10 media player, which is $699 -- anyone who pre-ordered the X10 before the 1st of this month got it for $499. Only catch is it needs HDCP 2.2 to operate. I've also heard it only works on Sony displays, but have not connected to the Samsung 4K in order to test that out. I know the Samsung movie-packs only work with Samsung TV's and are giving people all sorts of grief because most of the content supports 5.1 audio, but Samsung TV's notoriously have issues outputting better than 2.1 audio, even through HDMI ARC. Anyway, I'm rambling...
 
No Jeff, it's useful information to know. So you are saying that Redray is end of life, or just thus player. What you are saying about lack of HDMI 2 support etc, how many tines has Red not listened and we were right. They need to hire more people to cover all the bases.

About Redray, I think we should judge things on the basis of the 200 inch screen experience across our centre and center wide angle field of view, and should pixel peep on this basis. While I don't expect it to be lossless, like some acted like, I do expect it to be visually lossless without artifacts. While they can be seen, they shouldn't be there. So, Jeff, are these artifacts innocent bugs, or inherent to the algorithm at 36 mb/s? The question remains, how much more bit rate do you need in h265 to match Redray. I can further say that visually lossless is a bit if a cop out. I remember, before knowing anything much about these compression techniques, pixel peeping visually cineform footage years ago, fairly 2 dimensionally and what I saw did not impress my much. The reducing of contrast in details was disconcerting and looked a lot less appealing compared to the lossless original. When viewing from a small angle of view it may make less of a difference, but the true test is how striking to the viewers conscious and sub conscious mind it is on a propper large angle of view screen at distance, which might be twice as striking on some visually lossless systems for a scene. In the end, big budget, that is what film companies are looking for, and probably why top directors spend weeks examining camera footage before choosing which camera. So I have not been terribly impressed by s lot of footage I've seen from the cameras in the past either, because of olpf raw debate ring and less than lossless compression and hard color, that put it behind. I understand the face texture issue now, that really bugged me. Raw still looks less detailed and I forgot if misdetailed. Layered color sensors are probably going be the choice in future, and it looks like most major consumer manufacturers (including Canon) have designs for them.

All this content locking in the industry is monopolistic antitrust, and even if done overseas, it is being imposed on American consumers and violating American laws. Just because a person is overseas they are performing the action on American soil to American citizens, flouting American law. I've been thinking of raising this locally. It is not about communications law, it is about action across the border where you are in two or more jurisdictions at once (Red might like to keep that in mind for challenging monopolistic content lock out deals in future court cases).

'A life where you don't own things you have, but they are owned by others, is a definition of slavery. This applies to the control of things you have also.'

Now back to Sony Displays. I've looked through them last night and they still only go to 4k 8 bit p50. So I guess they are still stuck at HDMI 1/2 of 2.0. Now, if they were 4k 10 bit p50 I would buy one. They even saw one that still said HDMI 2 was by after market upgrade. With the amount of color, and consider the sheer price, they should have 12 bit modes to handle gradients. Pity they don't have a good interface like the latest Display port and Thunderbolt, and twin full speed HDMI 2.0 modes for 16 bit 4:4:4 p50, and rec2020 upgrade support. Even as a consumer display with that wide of a color gamut on high contrast I am worried about 8 bits.

Now Jeff, I know there is a funny color going through these displays, but in your opinion (as one of the few I trust around here) can the Sony's at all be balanced for professional use. Thanks
 
Why don't Red release Redcode as a free codec to encode and use for viewing material that is available to the general public? Can't shake of the feeling that Red Ray seem to go downhill that it's a kind of DOA product, marketwise?

If there was a codec that could seriously compete with H.265 and get a following then it will generate better solutions for us who shoot Red and also make people more aware about Red digital cinema.
 
I know what is likely happening. Compression codecs can be a field of licenses. Unless all licenses agree to a free release you are constrained from doing it Also freedom breaks profit from a codec. I agree that it is good to have free players, and to charge for encoders. This reduces revenue a lot, and codec manufacturers like to make customers pay three or more times at once (through codec use in encoder software player, hardware chips etc). Red could change all this by just charging on the encoding side, certainly player manufacturers in China would be happy, they hate paying the licensing dees on DVD players. Still there is another way, just charge the chip set/software player manufactures, and the encoder manufacturers. They could level the playing field. Imagine Google youtube with a Redray license, with Red getting a cut of the advertising revenue. The thing us VC, they always pull something else, even a version of Redray.

Another thing is that Red said they would tell about Red ray, but hasn't, the design is hidden away, and they want to keep it secret by looks of but, which again may have something to do with licensing.

The other thing is that they can distribute their own players and encoders linked to advertising. I know about these things, because this was my own plan for my software from the 90's.
 
Just to keep it real, it's worth noting that a big part of why .RED encoded material looks so good is that the specialized hardware (custom ASICs mostly AFAIK) in the player allows it to achieve real time playback from a far more complex codec than HEVC, VP9, etc. Even if RED was inclined to take the code open source, which I doubt they are, I would estimate 99% of computer hardware in the field would choke on it.

Cheers - #19
 
It should be simp!er :)

Opening it up allows companies like ati, nvidia and Intel etc, to augment their gpu's with circuites to assist decoding. RR, as they do with others. We assume that the rr hardware is so special it is far ahead of alternatives, I think it is not, just under central control. But to get speedy hardware you need to open it up for specialists like ATI and Nvidia, to do it. The same issues faced h264 etc for years until it was done. However, if their is a mismatch between RR and GPU dx10/11/12 level hardware (the commodity processing option at the moment) there are techniques to adjust it to be more executable on these GPU architecture. You land up with RR b, but at least it might work ten to a hundred times better per watt on a GPU at a fraction of the price. You have astarting point for expansion of the market, and to head off h266. If anybody wants to say this speculation is wrong, then release the specs and details so we can comment correctly. A lot of us put a lot of time and money in trustfully waiting for things to come through.At the moment it is not ideal, and less hopeful than before.
 
A lot of things "should" be simpler. Perhaps the engineers at ATI/nVidia/et al could harness the power of modern GPUs with an assist from a specialty chip on the same silicon to do what the RedRay does - maybe. Even if they could, would it be worth the cost to the GPU vendors to service a tiny fraction of their market? Hard to say with authority since so little is publicly known about the technical workings of the RedRay.

I also wonder just how "simple" these types of implementations are, at least from jump, based on the lead times we are seeing in developing commodity chips to decode HEVC/H.265 which is an exponentially bigger market with purchase orders already in the IN box.

Cheers - #19
 
Back
Top