[assassin]

Sorry, Back to the question at hand about Macroblocking...

I cannot answer as I am waiting on the i3 version for my personal use. We make Ivy i5s almost every week though and I can contact the shop to get some information for you on our next XBMC Ivy build.

[GreenEyez]

Try playing around with the contrast and brightness control in the XBMC OSD, Like assassin noted there seems to be a difference in them between the software and hardware pics, too low brightness can cause an aparent loss of detail

PS: liquidsnake, how did you manage to take the screenshots (last 2) on a perfect frame to frame match? I`m trying to do the same, but getting the right frame in a 24 frames / second sequence is a pain. Is there any skip by frame option in XBMC i am missing ?

[liquidskin76]

BTW thanks again for posting.

Not trying to be an %$#. Just trying to learn something different/new.

No problem!

Cheers for helping out on original question.

---

PS: liquidsnake, how did you manage to take the screenshots (last 2) on a perfect frame to frame match? I`m trying to do the same, but getting the right frame in a 24 frames / second sequence is a pain. Is there any skip by frame option in XBMC i am missing ?

I used the option to pause during refresh rate change, and set it to 20 seconds. Gives you plenty of time to grab a screenshot!

Cheers

[assassin]

Just heard back from an avs member. No dxva support with ivy using the latest nightly.

[liquidskin76]

Cheers for the info assassin. Really appreciate the help.

[nelson01]

I've used an hd 4000 and an hd 3000. The 4000 is much improved over the 3000 in regards to dxva macroblocking. It's about on par with the ati cards with regard to dxva decoding errors. In fact, what I've found with using dxva on ati and ivy bridge graphics is that the glitches during playback occur at the same point in the video and produce the same artifacts. I'm playing back untouched blu ray rips. As another user on these forums pointed out, these glitches only seem to occur on avc encoded streams with more than 3 reference frames. Oddly enough, if you skip back to before where the playback glitch occurred and let it play again, it doesn't happen a second time. This is usually the beginning of the video.

@liquidskin76
In the screenshots, what are you using as the render method for the software decode? I'm wondering if there will still be a difference if you decode with the cpu and use dxva to render the video.

[wsume99]

The 4000 is much improved over the 3000 in regards to dxva macroblocking.

Much improved? Are you sure you are talking about the same problem that is being discussed in this thread? With DXVA enabled on my G620 the video is totally unwatchable. We're not talking about decoding artifacts here. If you have HD4000 and you enable DXVA then I think you'd immediately be able to tell if the problem has been fixed. There may well be other issues with DXVA and Intel HD XXXX graphics but if you're able to watch the video at all then I'd say it's been fixed.

FWIW - there was/is a pre-Eden nightly build that was distributed (maybe 6-8 months ago) where DXVA did work with Intel SB graphics. I've confirmed this with my G620. I think may still have a copy but I'd have to check.

[nelson01]

The 4000 is much improved over the 3000 in regards to dxva macroblocking.

Much improved? Are you sure you are talking about the same problem that is being discussed in this thread? With DXVA enabled on my G620 the video is totally unwatchable. We're not talking about decoding artifacts here. If you have HD4000 and you enable DXVA then I think you'd immediately be able to tell if the problem has been fixed. There may well be other issues with DXVA and Intel HD XXXX graphics but if you're able to watch the video at all then I'd say it's been fixed.

FWIW - there was/is a pre-Eden nightly build that was distributed (maybe 6-8 months ago) where DXVA did work with Intel SB graphics. I've confirmed this with my G620. I think may still have a copy but I'd have to check.

I own both and have used both, and the hd 4000 IS much improved over sandy bridge graphics. Is it perfect? No, but it is much better than sandy bridge. FWIW NO gpu is perfect with dxva on the files I play. Not even nvidia cards play back everything perfectly. As I said before, the dxva on ivy bridge works about the same as ati gpus. The dxva playback I'm referring to is that of xbmc's dvdplayer. Playback with dsplayer and lav filters works better across the board, although cpu usage is slightly higher.

[wsume99]

I own both and have used both, and the hd 4000 IS much improved over sandy bridge graphics. Is it perfect? No, but it is much better than sandy bridge. FWIW NO gpu is perfect with dxva on the files I play. Not even nvidia cards play back everything perfectly. As I said before, the dxva on ivy bridge works about the same as ati gpus. The dxva playback I'm referring to is that of xbmc's dvdplayer. Playback with dsplayer and lav filters works better across the board, although cpu usage is slightly higher.

I'm trying to resolve why you are reporting that the macroblocking issue is much improved yet assassin is hearing reports that it is not. I see this XBMC Intel HDXXXX DXVA macroblocking issue as pretty much a black/white issue - it's either horrible/unwatchable (still broken) or it's barely noticeable (fixed). So I'm confused how anyone could be wrong yet we are apparently receiving opposite answers to this question.

[liquidskin76]

In fact, what I've found with using dxva on ati and ivy bridge graphics is that the glitches during playback occur at the same point in the video and produce the same artifacts. I'm playing back untouched blu ray rips. As another user on these forums pointed out, these glitches only seem to occur on avc encoded streams with more than 3 reference frames. Oddly enough, if you skip back to before where the playback glitch occurred and let it play again, it doesn't happen a second time. This is usually the beginning of the video.

Yeah, i've seen that to on untouched blu-ray backups, but i re-encode everything with x264/handbrake anyway, so don't see it on a day-to-day basis.

In the screenshots, what are you using as the render method for the software decode? I'm wondering if there will still be a difference if you decode with the cpu and use dxva to render the video.

On auto detect i believe. I'll check it out in a bit.

Cheers

[liquidskin76]

In the screenshots, what are you using as the render method for the software decode? I'm wondering if there will still be a difference if you decode with the cpu and use dxva to render the video.

Same no mater what render method i use. Increasing contrast and decreasing brightness via XBMC's OSD does remove some of the washed out look, however there's definately less detail with software decoding vs DXVA2. With software there's some smoothing and grain removal going on.

Cheers

---

Nice to see some more feedback regarding Ivy Bridge and macroblocking. I see the processors come with 2500 and 4000 graphics. Maybe improved with the 4000?

@assassin... what graphics was the AVS user running? 2500 or 4000?

Cheers

[nelson01]

Same no mater what render method i use. Increasing contrast and decreasing brightness via XBMC's OSD does remove some of the washed out look, however there's definately less detail with software decoding vs DXVA2. With software there's some smoothing and grain removal going on.

Thanks for trying that. I was really hoping it would be the same.