V7 Crashes while rendering with Cineform AVIs

DCV wrote on 9/13/2006, 9:08 AM
I don't know about anyone else, but I'm having some major problems today getting Vegas 7.0a to render anything using 1080-23.976P AVI files captured and converted using ConnectHD 3.04. The project is 720X480 Widescreen 24P. The same thing works fine from Vegas 6.0d.

I've tried frameserving out via Satish's frameserver as well as the MainConcept Encoder. Basically it appears there's a memory leak somewhere. Available memory counts down as the render progresses until Vegas crashes at zero with an exception that references the CFHD.dll.

I hope this can get fixed soon as Vegas 7 is dead in the water for me right now.

John

Comments

LJA wrote on 9/13/2006, 9:14 AM
Same thing happens when previewing CineForm AVIs from the time line. Page file usage increases until Vegas crashes. I have checked this on CF 2.5 and CF 3.0 files. Results are consistent. I am using CF HDV 720p intermediates.
Wolfgang S. wrote on 9/13/2006, 10:23 AM
That are bad news. The page file increase was the reason why Vegas 6 has not been able to hold more then 80 m2t files in the timeline. If that takes place now for Vegas 7, then we have an issue.

Desktop: PC AMD 3960X, 24x3,8 Mhz * GTX 3080 Ti * Blackmagic Extreme 4K 12G * QNAP Max8 10 Gb Lan * Blackmagic Pocket 6K/6K Pro, EVA1, FS7

Laptop: ProArt Studiobook 16 OLED (ProArt Studiobook 16 OLED (i9 12900H with i-GPU Iris XE, 32 GB Ram. Geforce RTX 3070 TI 8GB) with internal HDR preview on the laptop monitor. Blackmagic Ultrastudio 4K mini

HDR monitor: ProArt Monitor PA32 UCG, Atomos Sumo

DJPadre wrote on 9/13/2006, 10:34 AM
These SHOULD have been picked up with testing...
Product protection during development i understand and appreciate, but as you say, if these issues (which were first born from V6) take place now for V7, then there IS an issue..

there have been issues since quality control went out the window...
Wolfgang S. wrote on 9/13/2006, 10:52 AM
It is an illusion to believe that eveything can be picked up with testing.

Since the direction seems to be more to go toward native m2t editing, I assume that maybe nobody had the idea that there could be an issue with page file and cineform intermediates in the new version.

I assume further that it is now another issue. For V6 you had the page file issue for m2t files - and now you have that for cineform avi files.

Desktop: PC AMD 3960X, 24x3,8 Mhz * GTX 3080 Ti * Blackmagic Extreme 4K 12G * QNAP Max8 10 Gb Lan * Blackmagic Pocket 6K/6K Pro, EVA1, FS7

Laptop: ProArt Studiobook 16 OLED (ProArt Studiobook 16 OLED (i9 12900H with i-GPU Iris XE, 32 GB Ram. Geforce RTX 3070 TI 8GB) with internal HDR preview on the laptop monitor. Blackmagic Ultrastudio 4K mini

HDR monitor: ProArt Monitor PA32 UCG, Atomos Sumo

LJA wrote on 9/13/2006, 10:53 AM
I originally discovered this problem when comparing preview rates between m2ts and CF AVIs. The AVIs are significantly slower, slower (average 15 fps at "draft - full", but the more serious problem is the failure to release storage when reading CF files. Every combination of preview settings produces the same result. It even occurs with a single 30 sec. file if you preview it enough times. If anyone is interested a typical crash dump is below. I will submit this to Sony.

Sony Vegas 7.0 TRIAL
Version 7.0a (Build 115)
Exception 0xE06D7363 (Unknown Type) IP:0x77E73887
In Module 'kernel32.dll' at Address 0x77E60000 + 0x13887
Thread: VideoRender ID=0xE78 Stack=0x855D000-0x8560000
Registers:
EAX=0855dd90 CS=001b EIP=77e73887 EFLGS=00000206
EBX=00000000 SS=0023 ESP=0855dd8c EBP=0855dde0
ECX=00000000 DS=0023 ESI=0855de20 FS=003b
EDX=09430608 ES=0023 EDI=0855de20 GS=0000
Bytes at CS:EIP:
77E73887: 5E C9 C2 10 00 6A 18 68 ^....j.h
77E7388F: 28 30 E8 77 E8 40 6A 00 (0.w.@j.
Stack Dump:
0855DD8C: 097CA8A8 09770000 + 5A8A8 (CFHD.DLL)
0855DD90: E06D7363
0855DD94: 00000001
0855DD98: 00000000
0855DD9C: 77E73887 77E60000 + 13887 (kernel32.dll)
0855DDA0: 00000003
0855DDA4: 19930520 18EA0000 + A90520
0855DDA8: 0855DE30 08460000 + FDE30
0855DDAC: 097CB960 09770000 + 5B960 (CFHD.DLL)
0855DDB0: 0003A678 00030000 + A678
0855DDB4: 0855DE08 08460000 + FDE08
0855DDB8: 7FFA0BF8 7FFA0000 + BF8
0855DDBC: 00000000
0855DDC0: 00000000
0855DDC4: 01000E5C 00F80000 + 80E5C
0855DDC8: 00000000
> 0855DDDC: 77FA88F0 77F50000 + 588F0 (ntdll.dll)
> 0855DDE4: 097ACFB0 09770000 + 3CFB0 (CFHD.DLL)
0855DDE8: E06D7363
0855DDEC: 00000001
0855DDF0: 00000003
0855DDF4: 0855DE14 08460000 + FDE14
> 0855DE1C: 097CB960 09770000 + 5B960 (CFHD.DLL)
0855DE20: 0855DE34 08460000 + FDE34
> 0855DE24: 097B9BE9 09770000 + 49BE9 (CFHD.DLL)
0855DE28: 0855DE30 08460000 + FDE30
> 0855DE2C: 097CB960 09770000 + 5B960 (CFHD.DLL)
- - -
0855FFF0: 00000000
0855FFF4: 00590EC0 00400000 + 190EC0 (vegas70.exe)
0855FFF8: 046C5FF8 04480000 + 245FF8
0855FFFC: 00000000

Wolfgang S. wrote on 9/13/2006, 10:57 AM
"I originally discovered this problem when comparing preview rates between m2ts and CF AVIs. The AVIs are significantly slower, slower "

Yes, unfortunately true - playback from native m2t files is faster now, compared with Cineform Intermediates. But here Dan told us in another thread, that the newest update of connect HD will improve preview performance hopefully.

Desktop: PC AMD 3960X, 24x3,8 Mhz * GTX 3080 Ti * Blackmagic Extreme 4K 12G * QNAP Max8 10 Gb Lan * Blackmagic Pocket 6K/6K Pro, EVA1, FS7

Laptop: ProArt Studiobook 16 OLED (ProArt Studiobook 16 OLED (i9 12900H with i-GPU Iris XE, 32 GB Ram. Geforce RTX 3070 TI 8GB) with internal HDR preview on the laptop monitor. Blackmagic Ultrastudio 4K mini

HDR monitor: ProArt Monitor PA32 UCG, Atomos Sumo

SimonW wrote on 9/13/2006, 11:03 AM
Hmmm Odd one this. I've just completed a major project with Vegas 7 using the Cineform codec with a few hundred files. No crashes thus far. Completely at odds with my experience with Vegas 6!

DJPadre wrote on 9/13/2006, 11:04 AM
It is an illusion to believe that eveything can be picked up with testing.

((I never said i beleived all issues would be resolved through testing.. Im saying that a BASIC everyday work routine LIKE THIS shouldnt have these issues.. IF testing int he wealworld was done as we see here (ie a real world scenario of what if's) then the issue would most definately have been picked up))

Since the direction seems to be more to go toward native m2t editing, I assume that maybe nobody had the idea that there could be an issue with page file and cineform intermediates in the new version.
((Consideirng PF issues were rampant on the launch of V6, it seems that were back in square one.. and as for moving toward native M2t... maybe they should have considered using the already available technologies from various display manufacturers to tap into said HW to assist in performance and processing, as opposed to relyin on SW code which is obviously untested with THIS issue... ))

I assume further that it is now another issue. For V6 you had the page file issue for m2t files - and now you have that for cineform avi files

((And TIFF... and MOV... and anything else which doesnt use the Sony or MS codcs..

Go render a 96dpi 12mp JPG down to 76dpi for video 1080i (3mp equivalent) and tell me how the pagfile behaves...
THis is what i mean.. PF is there.. Mismanagement of it has ALWAYS been there.... working around it has always been a necesity...

Still though, its better than conforming everything before it gets to teh timeline.. . its the price we pay to have the ability to throw almost any codec on the timeline..

If only I could throw a divx file on the timeline.. Dont get me wrong.. of teh 100 people we hear from who are having issues, there are 1000 who are not having issues..
Marco. wrote on 9/13/2006, 11:10 AM
Similar here. I did many tests with Cineform AVIs, rendered HDV to Cineform, made playback tests. No problems at all.

Marco
DCV wrote on 9/13/2006, 11:15 AM
I also noticed a performance problem right away when working with Cineform AVIs in V7. It was sluggish for some reason and as I worked longer in the project, it got worse. I didn't notice the available memory decreasing until I went and rendered. The project is pretty small (20 mins, 25-30 clips). As I mentioned previously, 6.0d is working fine with same clips. My error details are much the same as LJA's. I wonder if it has anything to do with ConnectHD?

John
kcrosley wrote on 9/13/2006, 2:21 PM
Yep, I'm getting these kinds of rendering errors a lot with Vegas 7 (both full and "Movie Studio" versions). I work pretty much exclusively with Cineform compressed HDV files. Seems like these errors happen more or less at random for me though -- i.e., I'll get a crash one time and then can restart vegas and try again without incident.

Grumble.
kcrosley wrote on 9/13/2006, 2:22 PM
Has anyone tried the new Connect HD beta to see if that makes any difference with the Vegas 7 rendering issue?
Jayster wrote on 9/13/2006, 2:28 PM
The Cineform CTO mentioned today that if you are using HD Connect, you must uninstall/reinstall it when you setup a new version of Vegas (he said its a windows / VFW limitation). If you don't do this, Vegas 6 will continue to use the HD Connect supplied codec but Vegas 7 will use the Sony supplied codec.

Cineform is releasing an update of Connect HD in the next day or so. One of the features they are adding is multithreading of the VFW codec, which should improve playback speeds.

Hopefully the issue you guys are noticing will get fixed, too. (Or maybe it'll get fixed by simply reinstalling HD Connect, if you have it). But then again, maybe its a Vegas problem only, since it didn't happen in an out-of-box V6 install.

kcrosley wrote on 9/13/2006, 5:46 PM
Thanks for this tip, Jayster! I'll try it.

Cheers,
Keith
DCV wrote on 9/14/2006, 8:17 AM
Okay, I've been working this issue and here's an update:

1) I uninstalled Vegas 7/DVD Architect 4.

2) I also uninstalled Connect HD 3.04.

3) I reinstalled Vegas 7.0/DVD Architect 4.

4) I fired up Vegas 7, placed a Cineform AVI on the timeline, and played it. Available memory drops to the tune of 20-30 meg per second and paged memory increases as clip is being played. Eventually Vegas freezes up and crashes.

5) BUT, if you open the clip in the trimmer and play it, all is well. Memory does not leak and Vegas doesn't crash. Same is true for clip auto preview, works just fine. Copying portions of the clip from the trimmer to the timeline and playing from the timeline leaks memory as before and crashes Vegas.

5) I installed ConnectHD 3.2 Build 26 Beta that Dan posted to the forum.

6) The results in Vegas were the same, memory leaks when played from the timeline.

Are there any other people having issues with Cineform AVIs in Vegas 7?

John


fldave wrote on 9/14/2006, 8:57 AM
I tried Cineforms last night with no problem. Cineform V2.5 codec, though. Memory started rising when I started playing them, but stabalized after a while.
ECB wrote on 9/14/2006, 9:59 AM
I tired Cineform using the Cineform v2.5 codec with no problem. Memory usage rose slightly and remained flat. The same with page file usage.

Ed B
rmack350 wrote on 9/14/2006, 11:19 AM
Could this be related to the amount of Preview RAM being set aside in the prefs?

Vegas would probably be caching frames in order to play back the media. If Vegas has too much Preview RAM then things will start to get written to the page file. Might not even be Vegas going into the PF, could be anything that's in memory.

<Edit> I tried playing around with some rendered CF2.5 HDV1080i files (from noise textures) and didn't have any problems there. However, I did get some crashes on a full 1080i project file when rendering CF 2.5. Can't say why, but there was mixed media on the timeline (some DV files) and it wasn't an HDV cf template. Nothing conclusive, really.

The one thing I noticed, though, is that Vegas gave me a warning about page file usage when I tried to raise my RAM preview setting. Just a warning to say there could be consequences. I'm glad to see it!

</edit>

Rob Mack
kcrosley wrote on 9/14/2006, 12:38 PM
I can play Cineform AVIs on the timeline fine (though playback seems somewhat slow), but rendering (to any format) will often cause a crash.

With the release version of Cineform installed, these crashes caused Vegas to show one of its memory dumps. With build 25 beta of Cineform 3.2 installed, Vegas will hang during the render without showing an error message.

:(
kcrosley wrote on 9/14/2006, 12:52 PM
BTW, not sure if this is helpful or not, but the Cineform AVIs that I use are typically 1080 24p (captured with Connect HD)...

-K-
DCV wrote on 9/14/2006, 1:47 PM
My Cineform AVIs are 1080 24p (captured with Connect HD) as well.

John
LJA wrote on 9/16/2006, 9:41 AM
I have now tested the codec included in the Connect HD 3.2 beta build 28. Problem no longer exists, at least for me.