Posts: 1,388
Joined: Aug 2009
Reputation:
17
2012-07-22, 11:00
(This post was last modified: 2012-07-23, 01:15 by liquidskin76.)
Hi,
Does anyone know if the Intel Ivy Bridge suffers from the dxva macroblocking problem, like sandy bridge did?
Cheers
Posts: 479
Joined: Feb 2012
Reputation:
20
No difference in picture quality and no reason to require it's use either way.
Posts: 479
Joined: Feb 2012
Reputation:
20
2012-07-22, 23:45
(This post was last modified: 2012-07-22, 23:46 by assassin.)
It uses about 3 watts more on my sandy bridge system.
The softer pq is a new on on me. Source? By the way this is a xbmc problem and not an Intel problem. Maybe you should ask the xbmc devs to fix it.
Posts: 1,388
Joined: Aug 2009
Reputation:
17
right... ok, cheers for that.
Anyone else... does the Intel Ivy Bridge gpu suffers from the dxva macroblocking problem, like sandy bridge did? From what i've read, i believe not however would be grateful for confirmation based on experience.
Many thanks
Posts: 479
Joined: Feb 2012
Reputation:
20
Could you please provide the source that shows that dxva has better PQ? Maybe some screenshots? That would be appreciated since I have never heard of this before and it flies completely in the face of extensive testing I have done using NVidia, AMD/ATI and Intel iGPU.
I would be interested in seeing the results of this comparison while we are awaiting word on whether DXVA macroblocking has been fixed with IVB (as I said this is a XBMC issue and not Intel issue so I doubt its been "fixed" unless XBMC fixed it to be compatible with the Intel HDx000 iGPUs)
Posts: 479
Joined: Feb 2012
Reputation:
20
Can you post the same frame? Can't really compare the two otherwise as the lighting in that scene is completely different (look at the big guy's shirt in the background).
Thanks for posting btw. I still think the xbmc/ffmpeg developers could make it compatible with the current hardware.
Posts: 479
Joined: Feb 2012
Reputation:
20
2012-07-23, 01:22
(This post was last modified: 2012-07-23, 01:28 by assassin.)
I have seen a lot of screenshots before but this one has me perplexed. Why does the brightness (luma levels?) change?
After going back and forth between the two I actually prefer the software version as I think the other has far too much contrast.
Posts: 1,388
Joined: Aug 2009
Reputation:
17
2012-07-23, 01:30
(This post was last modified: 2012-07-23, 01:33 by liquidskin76.)
That's just the quality of decoding changing i guess? DXVA appears to be more vivid in colours, etc. On top of that, software appears to be smoothing and removing grain.
All post processing on catalyst control centre is off so no dxva post processing is going on.
I'm running both test on 24p refresh rate change by the way.
Posts: 479
Joined: Feb 2012
Reputation:
20
2012-07-23, 01:37
(This post was last modified: 2012-07-23, 01:39 by assassin.)
"Vivid" colors is completely subjective. Frankly I can't stand "vivid" or "torch" mode that many/some people like.
I can see where some people would like the picture that you prefer (DXVA) but I think that the software rendered version looks better to my eyes as its smoother around his face and shirt, for example.
In the end this is a completely subjective comparison between the two (and many other) options.
Have you tried MadVR with MPC-HC? That seems right up your ally. I can't discern enough of a difference to make me drop something with marginal (if any) PQ improvement if it means leaving something extremely easy to use like XBMC or Media Browser which would drastically decrease the WAF.
BTW I have tested MadVR and XBMC and I think the PQ is roughly the same.
BTW thanks again for posting.
Not trying to be an %$#. Just trying to learn something different/new.
