Lets give resolving the "flash frame" issue another shot.

Comments

Orcatek wrote on 3/31/2005, 5:43 AM
This flash frame- Does it only occur on the DVD? What I was wondering is if anyone had looked at the mpeg stream on the computer to see if it flashed there too.

Depending on what the result is one give Sony more to look at. If it does flash on the PC too, then they could analyze the file and maybe find something. If it is only on the DVD, then may once again the file could be looked at to see what is the issue the is causing the players to flash.

Also, has anyone checked to see if a flash DVD does it in all players or just some players. I know I've made DVDS that will freeze in some players and not others that I have had to fight with - I keep that old player around just for testing.



craftech wrote on 3/31/2005, 5:45 AM
No, it occurs when rendering to uncompressed Avi as well.

John
jetdv wrote on 3/31/2005, 6:03 AM
Set Vegas RAM preview to 0

Just be aware that setting it to zero can have a detrimental effect on render times. If you have a sequence that doesn't change (maybe a title over a still, for example) then it must render EVERY frame where if it is even slightly above zero it will zoom past that section using the first rendered frame.

Absolutely agree that this can't go on.

I'm sure that if someone can come up with a sequence: do steps 1 through 10 and you'll see the flash frame - then it would be fixed in short order. How to you fix a problem you can't duplicate? Plus, who's to say it's VEGAS that's causing the problem? It could be Vegas, it could be Windows, it could be the hardware, where do they look???

I too find it incredible that if Sony/Sonic Foundry had really spent a fraction of the time we have in watching hours of video footage to search for this glitch that they couldn't have reproduced it. It has happened to too many of us for me to believe they couldn't reproduce it themselves if they were motivated to do so.

John, let's say the HAVE seen it. Now how do you go about reproducing it on a development machine where you can really see what's going on? As you know, it's not simple to reproduce the problem.


For the record, I never noticed one in Vegas 3. I never noticed one in Vegas 4. I've only seen a couple in Vegas 5 and exiting and restarting caused it to go away. Plus, the ones I've seen were when PREVIEWING the timeline - not when rendering.
rmack350 wrote on 3/31/2005, 6:16 AM
Okay then. Yes, you CAN run renders over and over and you CAN compare two files bit for bit AND it wouldn't take a person to watch it. And it probably would be fairly fast to do the compare.

You don't even need to make sure any of the renders are faultless. You just render a master and then keep rendering over and over. All you need to determine is whether all the renders yield the exact same product. They should all be exactly the same, shouldn't they?

I've used shareware programs designed to hunt for duplicate image files. They did it by comparing name, size, checksum, etc or a combination of things you would choose. The checksum comparison would often turn up copies of images that differed only in name.

Rob Mack
craftech wrote on 3/31/2005, 6:22 AM
Plus, who's to say it's VEGAS that's causing the problem? It could be Vegas, it could be Windows, it could be the hardware, where do they look???
========
Give me a break Ed.

John
rmack350 wrote on 3/31/2005, 6:25 AM
It's a bit like playing the slots. You really hope the new version will add things that make it worthwhile. Vegas 5 involved a bit of personal bargaining for some to determine if there was something there that made the upgrade worthwhile. Could we live with the things that weren't addressed? Was there something appealing about the new version?

For me, there was enough in Vegas 5 to make the upgrade appealing but I can't recomend the program to anyone as long as this problem persists with no workaround. It really doesn't matter what new features are in V6. I can't recommend it to a new adopter.

Rob Mack
craftech wrote on 3/31/2005, 6:35 AM
This problem first appeared in Vegas 2 and has persisted ever since. Audio flatlines, ripple edit quirks, quantize to frames anomalies, amateurish credit roll function, not ready for prime time titler, blank frames, and inexplicable gaps are among the many others that have gone unaddressed as well.
The biggest improvement which was for me the last really siginificant one was the introduction of the excellent color correction tools in Vegas 4. That was worth it.

John
jetdv wrote on 3/31/2005, 6:37 AM
John, I'm just stating facts. I'm assuming there's probably a memory leak somewhere in Vegas causing the problem. But, even then, it's so ellusive how do you find it? And if it IS a memory problem, can you guarantee it isn't Windows causing the problem? Without steps to duplicate, you can bet it will remain - UNLESS it just happens to appear during testing (and is caught) on a develpment machine.

I've just had an Excalibur question sent to me about something not working. I got their VEG file, tried the same task, and it worked perfectly. Ok, now were do I start? Same thing here.
craftech wrote on 3/31/2005, 6:50 AM
Well let's see.
Sony assigns someone to try different versions of Vegas on different machines until the problem occurs. With so many people including DSE on a RELEASED dvd (according to what farss posted) having the problem it is bound to occur. When it does, they try to pinpoint it considering they have the codes, etc at their disposal. The impact it is having is more widespread than the example you posted. That is like comparing apples to sumack.
I would die if I ever handed a video to a customer with a glitch like that on it.

John
JJKizak wrote on 3/31/2005, 6:59 AM
The flash frame is unmistakeable when playing the timeline. However, I have rendered that small portion with the flash frame and did not see it in the render then when I played back the timeline it was gone.

JJK
craftech wrote on 3/31/2005, 7:02 AM
Sometimes I have played the timeline and didn't see it and it appeared after rendering. Are we getting any closer? Moreover, why are we the ones who have to figure this out anyway?

John
craftech wrote on 3/31/2005, 7:18 AM
craftech,

is it important enough to you to even try my earlier suggestion? what do you have to lose?
===========
I am sorry Marquat,
I should have told you above that I do plan to try your suggestion and any others I have not tried. And I do appreciate the suggestion very much.

John
craftech wrote on 3/31/2005, 7:35 AM
SonyEPM has posted this a few times:

"I've said it before: If anybody can provide a repeateable repro case, we can fix this. Send system specs and the .veg file to dr.dropout@sonypictures"

I have never understood how that would resolve the problem. It is unlikely a .veg file or system specs will get a repeatable glitch like this. One time he offered free software if anyone would send their hard drive to him. Maybe that would have worked, but no one did possibly because they still said it was contingent upon them being able to repeat the problem using your hard drive.

They need to devote the time to reproducing the problem on their own IMO. And I mean REALLY devoting the time to it since we didn't produce the software, they did.

John
johnmeyer wrote on 3/31/2005, 7:37 AM
I was able to reproduce a blank frame problem last fall and sent it to Sony. They said that it was the first time anyone had sent a reproduceable blank frame project. I received a reply from one of their developers.

They are well aware of the problem. It has to do with a "frame quantization problem which is hard to overcome in a mixed format/mixed project application like Vegas." The issue, as has been pointed out before, is that you have all sorts of video formats (NTSC, PAL, 30 fps, 29.976 fps, 24p, etc.) and all sorts of audio formats (I won't even attempt to enumerate) which need to line up perfectly on the timeline. In addition, if you have done any scripting, and try to write scripts to trim video clips, you will quickly realize that there are some amazingly complex issues of how all these things are stored internally. In particular, there are some very interesting roundoff issues (variations on the old high school problem of deciding whether 36.5 should be rounded to 37 or to 36). Making changes to keep the problem from happening in one circumstance may well cause it to reappear in another. The email I received from one of their engineers said that they were going to continue to try to eliminate the problem. However, I've had no reason to write up another bug report on the subject or correspond with anyone since last fall, so I don't know whether they have done anything for the next release.

[Edit]

Here are some suggestions offered in reply to the bug report I sent Sony last fall.

#3 sounded promising, but I will warn you that it is not practical in real life (way too many manual fixups required). Also, since it involves messing with a hidden setting, I would recommend that everyone skip it. I tried it, and it doesn't work.

For me, #4 was the key to avoiding the problem.
============================

Here is what I got from a Sony developer:

1) Use the delivery format's time format whenver possible- for DV or NTSC DVD use NTSC Drop (or non drop- no prob). You can actually use anything you want, but if you edit NTSC video in a PAL project and then render back to NTSC, you'd likely wind up with some wierd frame blending.

2) Enable quantize to frames- keep this on when editing video. You've heard this before I think.

3) You can view non-quantized video frames by ctrl+shift+options/prefs. Search on "quant" set both to true, restart app. If you'll open your blankframe.veg project, you'll now see bright red edges on the non-quantized event edges. It doesn't matter with audio, but for video it does, and in you project the edit point shows red at th cut. With quantize on, trim the right edge of event 1 to the left and the left edge of event 2 1 frame to the right, then but the clips together. Black is gone.

4) This probably happens to you most commonly when you throw a bunch of media on the timeline at once. Add event trim, add next event, trim, add next event, trim- far less likely.

craftech wrote on 3/31/2005, 7:42 AM
Maybe there is hope that they actually understand what many of us are looking for in this software. Unfortunately, you can "find" blank or black frames on the timeline. You cannot find the flash frames. Perhaps a different install for PAL and NTSC versions is in order.

Thanks for the post John.

John
johnmeyer wrote on 3/31/2005, 7:45 AM
Craftech,

Re-read my last post. I added a major edit after you posted.
craftech wrote on 3/31/2005, 7:48 AM
I did John and re-edited mine.

John
johnmeyer wrote on 3/31/2005, 7:52 AM
For those that want their very own "flash"/blank frame, here's the test case I sent to Sony. The blank frame happens where the two events touch.

The download is 2.8 MBytes. Click on the link below, and then click on the Blank Frame file to download:

Blank Frame Project File

You may need a valid Yahoo ID (like a Yahoo email account) to access these files.

John

Orcatek wrote on 3/31/2005, 7:57 AM
I wonder if the problem is linked to specific hardware - ie CPU's, chipsets etc?

Since we know AMD Cpu's are faster and somethings than Intel, maybe one of them is doing something that leads to the issue. Or it could be a combo of hardware - cpu, chipset, video card etc.

I've haven't had the problem yet, just thinking about possible reasons.

craftech wrote on 3/31/2005, 8:01 AM
John,
Are you setting your field order to "Progressive" as Marquat suggested above? How about "snapping" and "Deinterlace fields"

John
farss wrote on 3/31/2005, 12:44 PM
Well, firstly this is kind of interesting as it's the first time I've heard of this warning. Now I gotta agree with what they're saying about the dangers of mixing things at different frame rates and about editing with QTF turned OFF. And for goodness sake why do we have to drill down into hidden menu options to turn on a warning indicator? And why is being able to do what they claim is the very thing that brings us unstuck being used as a selling point?
Now oddly enough to the best of my knowledge this is the only NLE that'll let you do this (mix media at different fps), the fact that you can do it then isn't a feature it's a major oversight!

One big caveat to all this, it happens without mixing media and it happens without even making a single cut. I for one would never be so silly as to try to intercut NTSC and PAL on the one T/L, it is obvious to me that'd be a formula for something to go wrong just as I don't don't mix media at differnet playback rates etc. So from what I've seen this could be the cause of SOME of the problems people are having. I cannot see how a simple rounding error though would cause a frame from elsewhere in the project to turn up. You'd need a rounding error at say the minutes level not the frame level.
Bob.
rcampbel wrote on 3/31/2005, 12:55 PM
John,

I have only seen the flash frame problem once. I do think that it is different than a blank frame due to a quantization problem. The only "logical" way that a quantization problem would cause a flash would be to show something under the frame. But as I saw and others have reported the flash was from somewhere else on the timeline.

I just discovered the "show unquantized events edges in red option" myself a few weeks ago. I leave it on now since it makes it very easy to detect unquantized events. It is tedious to change a bunch of them manually, so I wrote a free script to do this last year.

The QuantizeToFrames script by default will quantize all selected video or audio events and will quantize the selected video event's fade times. You can change what is quantized by changing the value of a few vars in the beginning of the script.

Feel free to give it a shot to help with blank frame.

Randall
rmack350 wrote on 3/31/2005, 3:29 PM
There are some global prefs that should really be set at a project level. For instance (and I know I use this example a lot) if you set the default still image duration to 5.000 secs in a PAL project it'll need to be 5.005 for NTSC. So, it shouldn't be set in the global prefs.

Two different installs doesn't seem very appealing.

Rob Mack
musicvid10 wrote on 3/31/2005, 6:52 PM
Just to throw a little monkey wrench into the discussion:

I have W98SE, and the same anomaly occured with Pinnacle Studio MJPEG renders (albeit with a little more frequency) as well as with Vegas renders, most recently to DV and MPEG-2. I always thought it was a static (RAM) memory issue and still do, but you hardware types will have to figure out to your satisfaction if this is the case. Anyone have some insight if this occurs while conventional or virtual memory is being accessed?