Why poor quality PREVIEWING video through ATI AIW?

[haw]

I just bought an ATI All in Wonder Radeon 7500 video adaptor. I'm using the TV Viewer / Personal Video Recorder software to capture input from a VHS tape deck. The quality of the video is bad and I'm trying to figure out why ...

The scene I'm using to test my new ATI card is handled just fine using the same PC but different "preview hardware/software." I bought the AIW card thinking I would be upgrading my video capture capabilities ... not taking a step backwards.

I think this problem is related to the ATI's "TV software" DISPLAY capabilities more than the CAPTURE capabilities. The "bad quality video" I'm referencing shows up when I preview/watch the video in the ATI TV viewer, even before I capture it. And I see the same bad quality if I go ahead and capture the video, burn the digital video onto a DVD, and then play it on my set-top DVD player (on the TV). The problem is there at every step along the way (i.e., the CAPTURE process is working fine ... it's just that what's being captured is not good!).

I'll attach a couple of snapshots taken from the captured video, which illustrate the trouble. On the first image, below, note the "jagged lines" on the vertical bars of the iron fence, as well as the person's body.


On the second image, below, note the absence of these jagged lines. The overall quality of the image is far from perfect (I'm capturing from an old VHS tape, after all) ... but it is free from the "jaggeds" defect.



Any ideas on why this is happening or how I can correct it? I've tried playing with the ATI capture card's video soap settings, and other ones too, but none seem to make any difference.


My PC's hardware configuration is in my profile ... but the setup used to capture this video was:
1. Configuration that works (no "jaggeds"): VHS deck connected to Sony digital camcorder (where MPEG2 encoding takes place) which is then connected to PC via IEEE 1394 cable. Ulead VideoStudio7 used to capture (really just transfer) video.
2. Configuration that does not work (introduces "jaggeds"): VHS deck connected to "input head" that came with ATI AIW card, which in turn is connected to the AIW card. ATI TV 8.1 / Personal Video Recorder used to catpure the video.

Also, I have installed version 8.9 of the ATI MMC (even though the TV app says "8.1" on the splash screen) ... and 7.991 (6.14.10.6430) of the Radeon display driver.

[Manokat]

Looks like a field order problem.

Did you capture frame-based on UV7 and interlaced on ATI MMC TV? I bet ya.. you did. Try to capture deinterlaced on ATI TV you'll get the same result as UV7 frame-based.

[ajc53]

Welcome to the world of ATI.

Plenty of post and quides on this site addressing issues similar to yours and more.

ATI will probably be an upgrade in quality, but first be prepared to tweak hack and may be even buy additional equipment (higher end VCR, TBC, and lower end stabilizer) totaling about $750.

You can look for alternative capture cards but everything has a trade-off.
The most popular card on this site by far is the Canobus ADVC-100 for $250. But it captures DV AVI only, and must be encoded further for DVD use.

The biggest problem with ATI is it's built in Macrovision protection, mistaken less than perfect sources (which covers most old standard VHS tapes), and precedes to purposely screw up the signal. You can get all the technical details on this site. Just search ATI.

[haw]

Manokat:

ATI Capture was interlaced, field order B (though I cannot find where to choose the field order ... do you know)? All the Video Studio captures I did were also interlaced ... I experimented using field order A, field order B, and frame based ... none of which introduced the problem as I see it when using the ATI hardware/software. But I just did the capture step to get some "evidence" of the issue.

As I mentioned in my (probably too long!) original message ... this problem looks to me like it is being introducted by the TV **VIEWING** application ... BEFORE any capture is done. The problem is there even if I only preview the video and do not CAPTURE it at all.

I think it is the ATI viewer causing the problem because:
1. I can see the jagged lines when pre/viewing the video output from my VCR (without ever capturing it) in the ATI TV program.
2. I can pre/view the same scene in Video Studio with the analog signal run from my VCR through my digital camcorder and then into my PC via firewire ... and I DON'T see the jagged lines

Seems to me the camcorder is doing something different and better than the ATI display. But I don't know what. And I'm hoping there's a setting somewhere that I can change that will improve the ATI quality. Again ... if there is such a setting ... IT IS NOT A CAPTURE SETTING!


AJC53: I see (and have read) lots of posts regarding CAPTURING video ... I have not found anything that describes what I see in the TV viewer BEFORE any capturing is done. The ATI reference material did not help me. If you have any specific links to such info, it would surely be helpful to me.

[lordsmurf]

YOU ARE MERELY SEEING THE INTERLACED FIELDS.

This is fine. This is normal. This is proper.

The reason the other application did not show the interlace lines is because they playback/preview with a deinterlace filter. The other program capture you show has distinct traces of blended deinterlace in preview.

View this on the tv set. It will be fine.

Play your ATI file in PowerDVD or WinDVD. It will not show interlace lines because those players are equipped with bob deinterlace.

The ATI preview show show bob deinterlace too. This is before or during capture. After the file is captured, the interlace lines may show up in certain players.

Please visit my site and read the guide INTERLACE vs DEINTERLACE

The other possible answer is your video may have a "clear-water" effect (as I describe it) where you feel as if you're viewing the image through a shallow pond with ripples. This can only be corrected by a TBC, and your Sony camera likely has a TBC in it. This could ALSO be why one method sees an error and one does not.

I'd have to see a clip, not just a still capture, to determine the problem.

[Manokat]

So you get this even just previewing on MMC TV. You really need a TBC and or image stabilizer then. But before getting a TBC or IS try to preview and capture that same footage using UVS7 via ATI AIW capture card not via dv and firewire. If you get the same jagginess then you really need something to stabilize the video stream. Or just use analog to digital passthrough off the SONY dig. camcoder for VHS capturing and use ATI AIW for TV shows capturing.

[haw]

re: "view this on the TV set and it will be fine" ... I went ahead and burned the ATI captured MPEG file to DVD -- and even on the TV the jagged lines are still there. btw LS, I used your ATI capture guide to set my ATI AIW capture settings in the first place.

re: "play the ATI captured file in PowerDVD and it will be fine" ... I tried playing the captured MPEG file in PowerDVD ... still jagged. In both PowerDVD and when playing the DVD on my stand-alone DVD player/TV ... the picture looks just like it does in ATI MMC/TV preview (seems to me the capture part works fine ... it's capturing those jaggedies exactly as I see them!)

re: " so you get this even just previewing on MMC TV" ... YES!!! That's exactly right! The jaggeds are there whether I capture the video or not. If I DO capture the video then I have some evidence of the jaggeds ... but I can see them on my PC whether I capture or not. Are there any MMC TV settings that I can manipulate? I cannot find any setup options for MMC TV.

And speaking of setup options ... I see the ATI captured file is capturing "Field B" interlaced ... is there any way I can change this setting? In the Personal Video Recorder / Map Presets / Edit setting dialog, I see only the "de-interlacing options" drop-down ... which gives me a choice of captuirng as interlaced or de-interlaced ... but does not let me specify Field-A or Field-B. IS this setting found somewhere else ... or does it always capture field-b and there's no choice for me.

Finally, I'll try previewing and capturing using VS7 software with ATI hardware and see what happens. I'll let you know ...

[haw]

OK. Here are the results of my tests:
1. VCR > ATI input block > AIW card > MMC TV. Preview screen shows jaggeds. Captured clips show jaggeds when viewed on PC. Television shows jaggeds when clips are burned to DVD and played on set-top DVD player.
2. VCR > ATI input block > AIW card > VideoStudio7. Preview screen shows jaggeds. Captured clips show jaggeds. DVD/TV show jaggeds.
3. VCR > Camcorder > IEEE 1394 cable > VS7. Preview screen DOES NOT show jaggeds. Clips played on PC have no jaggeds. DVD/TV have no jaggeds.

My conclusion is still that the AIW card or MMC software is introducing this problem -- or maybe simply not removing it as well as the Camcorder/VS7 combination does. Either way, it looks like my investment of $100 in the ATI card was wasted. I would have been better off sticking with my existing camcorder/VS7 combination. I was trying to "upgrade" to the AIW card because I'm having some audio sync problems with the camcorder. I've done A LOT of research/experimenting trying to resolve that issue ... and finally decided that the camcorder was introducing the sync problems on those long 6 hour capture sessions. Since I don't have the sync problems with the ATI card, I guess I was right about that ... but the ATI card has its own problems, which are worse than the audio sync issue (that I can at least correct, even if it IS a pain in the butt)!

Well, enough of that. I'm frustrated by the whole experience. feel like I've wasted a month on the ATI card. Would appreciate any last ideas from anyone ... before I go throw the card out into the nearest freeway and start flaming ATI.

And although I appreciate the comments about buying a TBC and/or IS ... this is a personal endeavor for me, not a business proposition. I cannot spend another few hundred dollars on hardware to use just to convert these tapes I have.

Last thought. Maybe there's a better card out there for me that does a better job of CAPTURING video? I went ahead and bought the ATI AIW, thinking it would be better to buy more capability than I needed ... but clearly that's not the case and I have no plans to hook up television input to the card. I just want to be able to capture the analog input coming from my VCR ...

[Nelson37]

You've done several of the necessary steps, which if others would have actually read your posts would have saved some time.

Problem does NOT appear to be field order or de-interlacing, your TV playback results partially confirm this.

Suspect problem is with source which is being corrected by your camcorder.

Obvious next step. Use a different source. VHS has multiple inherent problems. Use either a DVD or cable, preferably digital. Even without the software Macrovision hacks, you should be able to cap 20-30 seconds from DVD. This should be enough to confirm capture quality.

Does your camcorder provide Analog out? This may allow using the onboard TBC while stil capping with the ATI.

[lordsmurf]

Fields are hard-coded in the device. ATI captures top field first. It is a constant, not a variable. Most devices are like this. For instance, DV is bottom field only, not changeable.

I have no doubt that the VHS tape is bad. Your Sony camera has a TBC in it that fixes the problem. This is a problem I see frequently on old tapes.

It is not the ATI card's fault, but rather poor source. That's video for you.

The card probably was an upgrade, but you forgot to upgrade the TBC aspect too.

[Nelson37]

No, field order is not hard-coded. The proof of this is that the field order changed somewhere between MMC 6.7 and 7.7, using the same card.

SFAIK, there is no software setting or even registry setting to change this.

[lordsmurf]

Field orders are too hard coded into the capture device (at least with THEATRE chips). The problem is older ATI MMC version did not properly create the files. So you have top-field files that think they are bottom field files. Depending on the MMC, some of the time you got bad files, other times you just got hard-to-work-with files

[Manokat]

I too do not believe that the field order is hard coded on theater chipset. Been using ATI AIW for a while now. Got 2 AIW 128 Pro, one AGP 4x version and the other AGP 2x version. I also have ATI AIW Rage Pro PCI (not 128) and AIW Rage II C+ PCI. Also have access to older AIW Pro 128 Pro PCI version and Radeon 7500 that are used at the church. Now I use AIW Radeon 9200.

Have bought and used few capping softwares with these AIW cards and can always pick field order I want, but not with MMC software.

IMO the field order is hard coded in MMC software not in ATI AIW hardware (chips).

[haw]

Nelson: I'll try capturing some other (better) source through the AIW card and see if it, too, has the problem with the jaggies (and report back). By now I'm sure, though, that the issue with THIS particular video is that it IS an old tape and the quality IS poor ... and that magnifies any inherent issues the ATI card (ANY digital encoding card) may have. AND ALSO ... yes, the camcorder does have another analog out plug (which I haven't been using)! Great idea ... sometimes the simplest things escape me!

All: another note for your regarding the field order question ... Even though MMC does not let one choose the field order ... Using the same AIW card, on the same PC, VS7 lets me choose between "field order A, field order B, and progressive" when setting up my capture. Of course, as I said, I cannot tell any difference in the captured clips between FO-A and FO-B. I'll also find some fast action source, capture a couple of clips both ways, and examine them to see if I can find a difference.

If any of you want to see these clips, I can ftp them to you. I've changed my profile to enable you to grab my email address (to send me an ftp address that you might not want to post) ... I'll leave it open like this for a few days and then I'll "turn it off" again. Alternatively, if you have any advice for me about specific things to look as I examine the clips myself, I can just relate the results to you here.

[Manokat]

Actually there is a "Fields" entry with default value "0" in the Windows registry. If i'm not mistaken it is in HKEY_CURRENT_USER in ATI section.

This entry is together with brightness, tint, color of input settings.

I thought the value "0" means field A. I changed it to "1" thinking that it may change to B. But no it does not follow that logic. Another thought is that "0" means disabled and "1" means enabled. But no, i ran mmc tv and try to check if there is new "field" option under video (brightness, tint, etc.). No success.

[haw]

OK. Here's my daily update. Thanks to all of you for hanging in there with me!

I captured some clips from a store-bought copy of Top Gun. The tape is in much better condition than the one I was originally using. And there are lots of good fast-action scenes. Here's what I figured out:

1. Regarding me not being able to see any difference between field order A and B clips. With more/faster action it was easy to see that the clips captured through my AIW card, using VS7 software, have the "flickering" problem indicitave of wrong field order when I specified field order A. So I guess the AIW card does indeed ouput field order B ... and that's why the MMC software doesn't let you see/set the FO. It would be nice if ATI actually documented this ... for those of us using other software to capture video using ATI hardware! And, fwiw, I guess ATI doesn't want you using their MMC software to capture video from other vendors' cards ... since you cannot change the field order setting (assuming the field order is dependent on the capture card hardware).

2. I don't really have the same problem with "jaggies" on the TG clips ... so I guess THAT particular problem lay with my poor-quality source tape.

3. I did notice one last problem, when doing my frame-by-frame analysis of the TG clips. Every fourth and fifth frame, ALL of the hard edges in the frame are blurred. In my case I was looking at airplane sillouettes against a blue sky (sometimes with white clouds ... usually not). The entire airplane silouette would have sort of a ghost image ... just a line or two above (when the plane was heading down in the scene). This blurring action repeated itself every fourth and fifth frame (stayed that way for two frames), with about three "clear" frames between. So the blurring happened six times per second.

I think this phenomenom has to do with my GOP settings ... I used the the 2P frames / 2 B frames recommended by Lord Smurf in his "Capturing MPEG with an ATI Card" guide. I also specified "closed GOP."

I am vaguely aware of the DEFINITIONS of I, B, and P frames ... but I'm not really sure if I should / how to go about adjusting them. How can I figure out WHICH type of frame the blurred ones are? Should I try to minimize these blurred frames or does your eye/brain overlook these when the final video is played? This blurring is a different thing than the jagged line problem which started this thread (so I guess I should start a new thread, huh?).

I guess these GOP settings will be my next area of experimentation ... so any words of wisdom from those of you who have been down this road before will be most appreciated!

[lordsmurf]

Good versions of ATI MMC grabs "TOP FIELD, A FIELD, 1ST FIELD" from ATI cards. All other programs I use or attempt to use must have that set if the capture is going to look good (if the software has an option).

No, altering the GOP will absolutely NOT give that reaction. What you are seeing is some form of deinterlace or improper capturing, likely caused by the software or a number of other factors. GOP has no bearing on the artifact you see. The other option is the TAPE blurs too. Yes, even film can blur (it's called slow shutter speed). The VCR can also cause this.

It sucks that you're having trouble, but you're far off base.

[haw]

LordSmurf: You say your ATI cards require field order A?!?! My little test indicates that my card requires field order B. I guess I won't worry about why your experience with ATI cards indicates that they produce FO-A video ... because I can see for myself, on my own hardware, that FO-B works and FO-A does not. But I think it's a little weird that my card does not seem to match up with your experience.

Then you say that "good version of MMC" software will output FO-A video? My MMC (said version 8.9 when I downloaded it a week or two ago from ATI's website) spits out FO-B video. I guess it's a good thing ... since my tests shows that FO-B video is what works for me, but I'm a little worried by your "good versions" comment (I sure don't want to be using a "bad version" -- after all, I AM experiencing some issues and I AM spending a lot of time trying to resolve them!). Maybe each ATI card DOES require a specific field order ... and the MMC software is able to query the card to determine the appropriate field order to use when capturing? Again, I'm not going to worry about this ... because I can see that my problem is not "using the wrong field order." But it seems interesting to me and I thought it might be interesting for you, also.


Now, regarding the "you are far off base" comment. That's why I threw out that idea ... I was hoping to get either some confirmation that I was on the right track or some comment such as yours to save me from going down the wrong path. Lord knows it's not the first (and won't be the last) time I've headed out in the wrong direction! So if it the problem is not related to GOP settings ... then I wonder what IS causing it?

I'll try doing some more captures ... using other VCR decks, other tapes, etc. Maybe through process of elimination I can eventually figure it out.

Meanwhile, let me ask another question. My first thought, when I saw the intermittent blurry frames, was that I was seeing RF interference of some sort. On some of the frames what I actually saw was a sort of "banding" where an entire band of horizontal lines would be blurred. These bands alternated up and down the entire height of my monitor ... so that I'd have a band of clear video (let's say 50 pixels tall) then a band of blurred (again, 50 pixels tall), then 50 clear, 50 blurred, etc. So ... I thought this looks like RF signal interference.

I DO have a wireless router (Linksys 2.4 Ghz "G") and a 5Ghz cordless phone system in my house. So to eliminate those items from suspicion, I unplugged them (from AC power) and captured some more clips. These clips still have the same banding / blurring issue so the problem is NOT tied to those two wireless devices.

But there MUST be an explanation ... the pattern and frequency of the blurred frames is just to consistent. Does THIS sound famiar to you or sound indicitatve of anything in particular?

If so, let me know. And if not, I'll be sure to post whatever I figure out...

[FOO]

You guys are confused. That top picture is NOT a field order
problem . It is horizontal sync jitter. Noise somewhere.

[Nelson37]

I am absolutely floored that none of the "experts' have identified your problem with every fourth and fifth frame. The video has been TELECINED. It consists of three progressive frames followed by two interlaced frames, the interlaced frames being a composite introduced in order to convert 24 fps video to 29 fps video. There are detailed explanations available, do a search. You might try the MMC IVTC function, "remove 3:2 pulldown."

As far as the ATI field order problem, I for one have made no effort whatsoever to determine what field order my file "thinks" it has. It could be hallucinating for all I care. It is absolutely clear that the field order REVERSED with no other change than the MMC version. I have seen this problem discussed in the past, with the same results observed. The card in question was a Rage 128, not sure which chip is onboard.

[lordsmurf]

This is a default MMC preset. Stupid one at that. Maybe it was left on? I forgot about that.

[next]

I am absolutely floored that none of the "experts' have identified your problem with every fourth and fifth frame. The video has been TELECINED. It consists of three progressive frames followed by two interlaced frames, the interlaced frames being a composite introduced in order to convert 24 fps video to 29 fps video. There are detailed explanations available, do a search. You might try the MMC IVTC function, "remove 3:2 pulldown."

As far as the ATI field order problem, I for one have made no effort whatsoever to determine what field order my file "thinks" it has. It could be hallucinating for all I care. It is absolutely clear that the field order REVERSED with no other change than the MMC version. I have seen this problem discussed in the past, with the same results observed. The card in question was a Rage 128, not sure which chip is onboard.

Thankyou

[FOO]

I wonder if you folks are aware that Ulead thinks Field B is the top field.

[lordsmurf]

I wonder if you folks are aware that Ulead thinks Field B is the top field.

Short term memory sometimes. I do remember that.

[Manokat]

FOO wrote:
"I wonder if you folks are aware that Ulead thinks Field B is the top field."

Absolutely. But they fixed that on Ulead Movie Factory 3 Disc Creator. They even use the name "Upper Field First" instead of Field A (like in UVS 7).

[haw]

For the video captured via MMC, it was absolutely positively captured WITHOUT the "inverse 3:2 pulldown" option enabled. I used my own preset when capturing the video, and that option is not enabled in the preset.

It is interesting that Nelson described the option as "REMOVE inverse 3:2 pulldown" ... that is NOT the way the option appears in my MMC (I'm now on version 9.0). My version say simply "inverse 3:2 pulldown." I do not have the check mark beside it. Since the checkmark is not there does that mean "remove it?"

At any rate, now that I know that option might cause the effect I'm seeing, I'll go back and do some tests with it enabled/disabled.


Researching outside this forum, I also found background information about the confusion of Field Order "labels." So I concluded the thing to do is not worry too much about it's called ... just do the test burns with each new hardware / capture software component and see which setting works, then use it.

[Nelson37]

3:2 pulldown is the same thing as Telecining. You are correct, the option in MMC is Inverse 3:2 pulldown, which may remove the problem.

Telecine can be done via software using an embedded flag, as on most DVD's, or hard-coded, recorded, that way. This is what you have on the VHS tape.

You can also remove Telecining with a software filter, but this will require re-encoding.