Little confused about huffyuv

[swiego]

As I try to learn this whole process, I've been having problems with interlacing and I think they are traced to my not understanding huffyuv fully. Assuming standard NTSC SD capture (480i from SVHS for example) to huffyuv,

1. Should my huffyuv-encoded AVI files be listed as interlaced or not, in GSpot? Mine are listed as 29.971 FPS HFYU (all nice and normal) but the little I/L and PROG indicators are off, as are the IFF/BFF/etc indicators. I realize that the source material itself is interlaced. Does the huffyuv format not specify this, or that it is TFF vs. BFF vs. ? Or am I not capturing correctly?

2. What settings should I be using? I have read digitalfaq.com, plus several sites and guides, and the advice is inconsistent. I keep the compression methods standard (Predict gradient, Predict Left) and have the Field Threshold at 288 lines. I also check "always suggest RGB format for output" and "Enable full size output buffer" and use v2.1.1. I'm very nervous about my Field Threshold setting, I don't think it's right but I don't fully understand what it is supposed to be.

3. Since TFF/BFF seems to matter so much when encoding to another format (e.g. mpeg via CCE or procoder) how am I supposed to know what settings to use in those programs when converting my huffyuv source?

Any help would be appreciated. I have gotten the results I want, but it's been hit-or-miss and less due to understanding what's really going on!

[RoopeT]

Since you're capturing NTSC you should use field treshold of 240 lines, 288 is for PAL.

Huffuv is lossless, so you should qualitywise get the same results regardless of your compression setting. Some setting require more CPU power and may compress better.

MiniDV sources are BFF, all other video sources are generally TFF.

[jagabo]

HuffYUV encodes interlaced and progressive frames exactly the same. So it doesn't need to know if the frames are interlaced or progressive. The "swap fields" option is only there to correct a bug where some old cards captured with the scanlines swapped -- it isn't to be used to specify field order.

You do need to be sure the program that receives the output from HuffYUV understands the video is interlaced and which field is first. You do this by setting those options in that program.

I wouldn't use the "always suggest RGB..." option because programs that can handle the YUY2 output from HuffYUV will work better (faster, more accurate) that way. Only use the RGB option if you have problems.

[swiego]

Thanks!

"All other video sources are generally TFF" --- this will sound like a silly question, but how can I tell whether it's TFF or BFF? I just tried moving frame by frame in virtualdub passing my huffyuv file through avisynth with AssumeTFF() vs. AssumeBFF() and saw no difference. Then again my capture was with field threshold set to 288 lines so my capture file could be completely messed up... I'll recapture and try again.

[RoopeT]

The field treshold setting is likely only used in combination with that "swap fields" option.

You can check the field order with TMPGEnc (encoder, demo will do if you don't have it):

Open your source video, set video source as interlaced 29.97 fps 720x480. On the "Advanced" section choose top field first & video arrangement method full screen. Enable de-interlace filter with even-odd field as deinterlacing method.

Click the intelacing filter and it will open the video for you. Watch your video by moving the slider. If your video playback seems smooth you have the right field order, if you notice any jerkyness change the field order as BFF and try again

[swiego]

The field treshold setting is likely only used in combination with that "swap fields" option.

Meaning that the setting doesn't or shouldn't matter if "swap fields" is off? (FWIW that makes sense to me.)

[RoopeT]

Shouldn't matter if swap fields is off. Anyway, the right value for NTSC is 240.

[jagabo]

Put the following AVISynth script in VirtualDubMod's Template folder:

#ASYNTHER TestTFF
[DirectShowSource("%f")]
AssumeTFF()
BOB()

Then open the the source file with VirtualDubMod using that template (ie, select the template via the little "Use AVISynth Template" pulldown at the bottom of VirtualDubMod's Open Video File dialog).

Then step through a part of the video where there's motion. If the motion moves smoothly the file is TFF. If the motion jerks back and forth it's BFF. If you see each frame twice, with no back and forth jerks, the video is progressive.

[thecoalman]

Although this is not 100% it is 99.9% correct... If the device you used to capture the footage is a DV device (DV Cam, Canopus ADVC) the field order is Bottom... other devices like capture cards are top..

There's a more thorough explanation here: https://forum.videohelp.com/viewtopic.php?t=257631&highlight=

[mikiem]

In case it helps, I think codecs with a fields over "xxx" or threshold setting use this as a switch, & send info to a requesting app accordingly on decode -- as in anything over whatever amount of lines is interlaced. Avi as a container doesn't really care or store info on fields, so this really is optional, and turning this on can mess up, even crash some editing software. But as everything else, set it according to what works with your software.

[swiego]

Well, I switched to lagarith and am having a *much* better time now... it seems to compress at least as well, if not better, and for better or for worse, it doesn't present me with options that may or may not make things worse. Somehow I can appreciate that