Capturing analog Video8 tapes comparison?
Hello,
I have been dealing with capturing analog 8mm tapes and I'm trying to figure out which capture chain (and player) yields the best results. Here are 2 clips which all show the same sequence. One has been captured by a professional transfer service right into DV-AVI (player, setup, capture chain is unknown, however, MediaInfo lists a DVCPRO codec for the respective DV-AVI files). The other file shows my own attempt of capturing. I'm using a Sony EV-S9000E (PAL) Video/Hi8 Deck, connected via S-Video to a Panasonic EH 575 which is then connected via HDMI to a Blackmagic Intensity 4K Pro Capture Card capturing in 4:2:2 HuffYuf lossless AVI --> I'm following this guide here in terms of capture setup because it was said to yield the best possible results for analog capture. Now here is the comparison: one by a professional transfer service done into DV-AVI with unknown capture gear and setup and the other one done by myself with the above-mentioned gear and capture chain. What differences do you see? Which would you think of as "better"? (I used an external hoster since the forum has a limit of only 99MB, which, using the original unaltered clips without re-encoding, is too little even for the DV-files that are by technical standard more compressed than the HuffYuf ones): Sequence 1 (Transfer Service): https://we.tl/t-jPxxdJVzrR Sequence 1 (own capture): https://we.tl/t-PaMCxJLohr Sequence 2 (Transfer Service): https://we.tl/t-C0F6s8PaWl Sequence 2 (own capture): https://we.tl/t-aiVLwAsUEk I can see quite some differences between the two different formats and capture chains. What do you see? |
Sorry but all is wrong... Why to HDMI? Why to HD and not SD?
|
Quote:
|
The DV files look to be de-interlaced, chroma suffered the most, For better comparison you should have de-interlaced your files. I see a lot of line jitter, Is the deck's line TBC on?
|
Quote:
|
Quote:
If I play them using VLC Media Player they look horribly jaggy because they don't have a proper interlace flag implemented for some reason. Same thing about aspect ratio. The Huffyuv-Files seem to have an aspect ratio flag saying 5:4 wheras the DV-files correctly have an aspect ratio flag of 4:3. Something about these flags must have gone wrong during capture - but you can easily tell your editing software how to "interpret the files". Also, de-interlacing analog material is never a good idea. Analog material is always interlaced - so it's best to keep it as it is. Modern TVs or players can easily handle it by deinterlacing themselves. No need to mess up the material by forcing deinterlacing. |
Quote:
|
QTGMC interlacing is fine, I don't trust hardware de-interlacing even the $3,000 LG OLED I have, it does a horrible job at interlaced videos, not bad for 1080i, but 480i de-interlacing is just an after though for almost every device made in the last 10 years or so. Watch interlaced only if you have an old school CRT TV if you can live with the flicker.
|
1 Attachment(s)
Hmm. Something odd seems to be going on with both, or it could be my playback hardware just using VLC. It's as though every other line is a slightly different color, like sky is blue with slightly more purple line in a repeating pattern. In times of no motion, you shouldn't really be able to see any sort of every other line patterning and even so, it should only be in the areas where there are motion and not static colors like the sky. See attached screenshot of this still relatively still part and check out the sky. This is of your capture, but I also see it in the professional capture as well. Could be the way the tape was recorded maybe where the odd fields somehow got recorded at different color levels?
|
Quote:
I understand that there's an Avisynth script to make them go away, though. |
I think he is talking about the entire frame tone, One field is greenish, the other field is blueish, I noticed that on the lossless sample only, The DV file has identical fields' tone.
|
Quote:
Color is off with both workflows, work is needed if that's something you're doing for these. Quote:
- If the footage is super sentimental, rare, critically important, and must be kept as highest possible quality, then your new captures plus editing will be better. - If the footage is "I'm using this sweet camcorder, I'm zooming in and out," then I'd take the previous captures and be done, and that would be better. It depends how you factor in all your time spent to recapture, for it to be a tiny bit more detailed (which no one will notice given the overall quality), then all your time spent doing additional work on the new captures. Everyone has different valuations for this. Quote:
Regardless of what AI upscaling and other goodies the AppleTV and Nvidia Shield currently have in their toolbox, they're not matching manual SD preparation quite yet. So, yes, any device "can easily handle" deinterlacing on its own. But there's still a visible quality difference in most cases. No one is saying deinterlace the archival masters. Just the files for streaming, sharing, etc. |
Quote:
Here is a frame-by-frame comparison between the DV capture and my own DIY-HuffYUV-capture: https://imgsli.com/MjQ0ODU5 While I cannot tell differences with regard to better details, that is, the 5-10% improvement that you are seeing, what I see is that the DV seems to have a little lesser brightness or gain which results in more fine structures that are preserved (e.g. cloud outlines in the sky). My own capture seems to be a tad over-exposed causing fine structures and details to be washed away in the sky (clouds). In this scene here it all becomes even more apparent: https://imgsli.com/MjQ0ODY2 |
DV brightness gain is more accurate than the lossless capture, This is one of the reasons why BM cards are not recommended for analog tape sources, You can tone down the gain in the crossbar, Also the sharpness created ringing and dark edges. DV suffers from chroma as I mentioned above and it is pumped up a little bit.
|
Quote:
|
Site design, images and content © 2002-2024 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2024 Jelsoft Enterprises Ltd.