Welcome to our community

Be a part of something great, join today!

Register Log in
  • 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 :)

IBC - GPU Build REDCINE-X Beta 21

Danai, could you post some transcoding times with your 3930K and 690? I have the 3770K and a Titan and am considering going to the 4930K (new mobo too obviously) and wondering how much of a difference the 2 extra cores actually makes. Right now I'm getting the same realtime playback performance you're getting. 1/2 debayer at 4K. I get 1/4 debayer at 5K. Thanks!
 
Grant Limberg said:
Thanks for the report, Les. We'll check it out!
Click to expand...

Hello Grant! Anytime soon REDCINE X will be able to export a copy or replace of the RMD/R3D file with the new sync external audio synced inside?
As early as last version it was not possible according to RED Squad Team
 
Anyone from Red, I imagine you guys have multiple varying systems you're testing with. What has been your experience on what is more beneficial for RedCine-X Pro with GPU acceleration and handling 6K R3Ds... more cores or higher clock speed? I'm deciding the proper course for my computer and wondering if I go the dual Xeon route which of these would be better:

24 cores @ 2.7 GHz
20 cores @ 3.0 GHz
16 cores @ 3.4 GHz
 
More cores should give you more than the higher GHz, but we've not extensively tested that. Four more cores (or 8) can do quite a bit more work, whereas frequency increase doesn't give that much extra fps (but it does increase).

If anyone has identical systems with CPU setups like Mike lists it will be interesting to see how the chips fall...
 
jacob.schwarz said:
Has anyone posted Dragon R3D stills for us to play with?
Click to expand...

I don't know where I am in the queue, but the first few days I have a Dragon, or even perhaps the next time I'm around one I'll shoot some material and upload some .R3Ds.

Might share little 3 second clips like I've done in the past so you can see it in motion.
 
Phil Holland said:
I don't know where I am in the queue, but the first few days I have a Dragon, or even perhaps the next time I'm around one I'll shoot some material and upload some .R3Ds.

Might share little 3 second clips like I've done in the past so you can see it in motion.
Click to expand...

What would be great phil! Thanks for thinking of us.
 
Hi Guys, now that you have probably looked at this more closely could you say if there will be a fix soon ?
Red is all about the image quality, I'm sure this little math error or round-off error or whatever it is will be rectified.
Thanks ! I'm really liking the speed, that's for sure.

Les Dittert said:
Here is a 400% blowup in a jpeg.

GPU_difference.jpg
Click to expand...
 
Les Dittert said:
Hi Guys, now that you have probably looked at this more closely could you say if there will be a fix soon ?
Red is all about the image quality, I'm sure this little math error or round-off error or whatever it is will be rectified.
Thanks ! I'm really liking the speed, that's for sure.
Click to expand...

That looks like a major deviation, and noticeable on sharp edges too. However, I am not able to replicate your results (GTX 780). Curious to know what conditions cause this behaviour.

Edit: After running some tests, I have finally been able to replicate the pattern in some situations. Had to blow up to 800% nearest-neighbour. It seems most common in areas around high contrast edges. What seems to be happening is that in a row of pixels, the contrasting colour tones are alternating instead of being a smooth linear gradient. Over multiple rows it looks like a checkerboard pattern.
 
this is probably less of a concern, as this is an old card, but tested this on a 4 year old notebook using a Nvidia Geforce GT130M (using the latest driver).

RCX 21 Beta does not debayer to FULL resolution, only displays a black screen - 1/2 | 1/4 | 1/8 work fine. Same thing happens when exporting a snapshot to 16-bit tiff, only creates a black frame.

Exporting a R3D trim works correctly though.

I sent the log file.

- M
 
I can confirm that what Les is talking about exists on my Titan as well. There also appears to be a one pixel shift in the x axis between identical still sequence exports and still exports.

Although difficult to see at 100%:
rcxGPUBug100p.jpg


At 400% you can see it for sure:
rcxGPUBug400p.jpg


From my initial observations it doesn't appear to happen everywhere in the frame, but only within certain values that roll to brighter values perhaps? I could be crazy. I'm still doing tests.

Running out a difference pass shows the variation between the versions fairly clearly.
 
I finally got around to testing this out (OC-d i7 3930, 64gig Ram, GTX 680 4gbs)
Man, does footage fly!
 
Thanks Phil,

I'm working on it .. we should have an update out in the next few days.

Thanks,
David

Phil Holland said:
I can confirm that what Les is talking about exists on my Titan as well. There also appears to be a one pixel shift in the x axis between identical still sequence exports and still exports.

Although difficult to see at 100%:
rcxGPUBug100p.jpg


At 400% you can see it for sure:
rcxGPUBug400p.jpg


From my initial observations it doesn't appear to happen everywhere in the frame, but only within certain values that roll to brighter values perhaps? I could be crazy. I'm still doing tests.

Running out a difference pass shows the variation between the versions fairly clearly.
Click to expand...
 
Sven: I assume you mean there is *no* audio after trim? That's correct, currently audio is removed. We'll get that sorted out somewhat soon.
 
You must log in or register to reply here.
Back
Top