X2 AVCHD SmartRender problems

Efenstor · Post by **Efenstor** » Thu Sep 18, 2008 6:35 pm

Ken Berry wrote:I am sure etech will offer his own views, but for me personally, I never use either SmartRender or two pass encode for either HDV or AVCHD...

I.e. you prefer degraded quality and hours of waiting? Interesting.

Post by **Ken Berry** » Thu Sep 18, 2008 9:35 pm

The hours of waiting never worried me. And believe me, the quality issue is something I am a stickler about. Maintaining the original high quality settings, a single render is not going to produce any degradation in quality which is detectable to the naked eye -- or at least my naked eye. And I am happier with that situation than I am with the hassles that occurred when I tried initially to use either SmartRender or 2 pass encode with AVCHD...

I don't know if it is a bug in the program in the true sense of the word, or just the nature of the AVCHD beast. But each to his own workflow!

Efenstor · Post by **Efenstor** » Fri Sep 19, 2008 8:12 am

It is certainly a flaw, because I tried to decode the produced file with DGAVCDec and it shows heavy artifacts exactly in the problemous places.

The reason of this problem is, I guess, that MPEG Transport Stream container is not well suited for AVC content, since it was originally created for MPEG-2. I think the story is identical to that with the AVI container, which in time gathered so much incredibly stupid twists inside that a rare player can play any kind of it correctly, because each app producing AVI's without using the Video for Windows SDK produces them in its own fasion thus sometimes making them very incompatible. Just read the swear comments Avery Lee left in the AVI reading module of the source code of VirtualDub!

I've already contacted Corel regarding that issue, I don't think they'll reply, but still it worth trying because Canon HF10/100 camcorders now rapidly becoming popular and for sure soon there'll be more complaints.

Post by **Ken Berry** » Fri Sep 19, 2008 8:31 am

I hate to admit it, but I agree with all your points, and particularly about the increasing popularity of not only Canon, but other AVCHD cameras as well. The recent emergence of the higher bitrates in the Canon range in particular will be a headache since VS doesn't seem to be able to handle them at all, at least going on the 24 mbps sample they have on the Canon website...

I suspect also that none of the software developers have thought hard about SmartRender since it was developed quite a few years ago now for standard definition DV and mpeg-2. In theory, the same principle should be applicable to other formats in general, and in particular to high def video. But I suspect that little or no effort has been made to match up the codes. Ditto with two pass encodes.

Just lucky I am patient, I guess!!

michal_h · Post by **michal_h** » Tue Oct 07, 2008 8:23 am

Hello,
I have the same problem as Efenstor wrote. I tried to disable smart renderig or cancel mpeg optimalization but nothing help

.
I use CVS12 x2 pro, camcorder canon hf100. Working with VS12 is excelent using smart proxy but output files (avchd pal hd 1920) have still mistakes.
I don't know where problem could be...(probably some bug in videostudio?)

bankroot · Post by **bankroot** » Tue Oct 07, 2008 1:14 pm

you've got worse spec of PC than mine has.
How is it possible, that you have no problem with editing AVCHD footage ???
In what mode (coding speed) do you record ?

michal_h · Post by **michal_h** » Tue Oct 07, 2008 1:55 pm

bankroot wrote:

you've got worse spec of PC than mine has.
How is it possible, that you have no problem with editing AVCHD footage ???
In what mode (coding speed) do you record ?

the best mode (1920x1080@50i,17Mbps), do you use smart proxy?

erdna · Post by **erdna** » Tue Oct 07, 2008 2:17 pm

Ulead/Corel's Smart Render is not up to a level that it can cope with the complex AVCHD codec implementations. Depending on the (in- camcorder) compression choices (GOP rate status, number of referance fields in use...) and the actual picture content at the very image changes the smart render algorithm gets lost resulting in all kinds of artifacts. By not using smart render, the codec can keep its own "game"and content predictions. Of course it is time consuming and needless(?) re-encoding but garantees smooth video. Maybe making the proxy approach more intelligen, by not allowing cuts on "difficult" frames would solve the problem

michal_h · Post by **michal_h** » Tue Oct 07, 2008 2:38 pm

erdna wrote:Maybe making the proxy approach more intelligen, by not allowing cuts on "difficult" frames would solve the problem

do you have any idea how to easy "not allowing cuts on "difficult" frames"?

erdna · Post by **erdna** » Tue Oct 07, 2008 3:16 pm

During the proxy generation proces, each video segment should be analysed w.r.t. preceding segment ending part, and compared to the next segment start and decide to skip the "incompatible" frames. This would of course make the intermediate codec process more complex (and slower) and sometimes result in lost frames.

michal_h · Post by **michal_h** » Tue Oct 07, 2008 5:35 pm

erdna wrote:During the proxy generation process, each video segment should be analysed w.r.t. preceding segment ending part, and compared to the next segment start and decide to skip the "incompatible" frames. This would of course make the intermediate codec process more complex (and slower) and sometimes result in lost frames.

lost frames are not my example, I have problem as Efenstor wrote: "portions of video taken a bit forward or backward in time, so that it looks either as a small repetition or a snap". In my case I have always backward in time.

michal_h · Post by **michal_h** » Wed Oct 08, 2008 9:50 am

to Efenstor:
have you already solved the problem? or you still waiting for new patch?

michal_h · Post by **michal_h** » Wed Oct 08, 2008 10:42 am

I've made a little observation and found the same mistake: one frame backward in time. I have three clips from canon hf100 (AVCHD, 17Mbps, 1920x1080@50i), put them on the timeline, used transition crossfade between clips, used some audio file from X2's offer, created video file AVCHD: PAL HD 1920. I made four outputs with different settings (I changed only these settings):

1) mpeg optimizer accepted, smart render disabled:
http://rapidshare.com/files/151983789/m ... r.mpg.html

2) mpeg optimizer accepted, smart render performed:
http://rapidshare.com/files/151985187/m ... r.mpg.html

3) mpeg optimizer cancelled, smart render performed
http://rapidshare.com/files/151986483/s ... r.mpg.html

4) mpeg optimizer cancelled, smart render disabled:
http://rapidshare.com/files/151987368/nothing.mpg.html

You can see in all outputs the same mistake in the same time:
00:04:10
00:05:12
00:09:07
00:10:18
00:16:11

I like this soft but this is really big problem for me. Have someone any idea how to fix it?
Thanks Michal

erdna · Post by **erdna** » Wed Oct 08, 2008 12:19 pm

I get the same results at the same times as you experienced. The hick seems to happen at the the crossover starts and end, and before the video ending. Did you also try the edit with the Proxy disabled ?

michal_h · Post by **michal_h** » Wed Oct 08, 2008 1:18 pm

I tried to create file while smart proxy is disabled but have still the same bad result, the mistake is always at the the trasition start and end

There is no mistake if I create file without transition

Here are two files from my camcorder, would someone by so kind to test it in his/her CVS pro X2? (put the clips on timeline, use any transition, create file AVCHD PAL HD 1920 and see the result)

http://rapidshare.com/files/152026040/00003.MTS.html
http://rapidshare.com/files/152026427/00005.MTS.html