1 Attachment(s)
Hi all,
I am involved in a project requiring the capture, restoration and editing of some PAL VHS tapes. I have 4 VHS tapes that run about 2 hours each, and which I have had a colleague of mine recently convert into ProResLT .mov (native PAL resolution, interlaced). From what I understand, the pathway was a JVC high end VHS deck, using S-video connection into a Datavideo frame sync and then S-Video into a Matrox card and then using some software called InstaCapture on a PC to record the 720 x 576 capture. From these raw captures, I intend to render out h264 mp4 files (interlacing intact) and send them to the good folks here at digitalfaq to process (minimise chroma noise, lines and other VHS artefacts..) (I need to send files via web, as I live outside of the US, hence needing to make the files smaller for web transfer). I have reason to believe that this test render I have done is not quite right, and was hoping to get a second opinion from a forum member here. Please see file attached. Thanks in advance for your time. |
The VCR used for this capture added some "ghosting" the bright white rim you see on the contour on the right of the players,
but, overall the picture is a bit on the dark side, (gamma correction?)i guess some of it can be corrected @digitalfaq, or you could try it yourself with Davinci Resolve which is mainly a color correcting/grading video editor, which i find a joy to work with, the easy way is to pipet the white and black levels in a frame, you can fine adjust these leves aslo if needed. The ghosting effect lies not in my field to adjust, you should ask others for that, i would try to capture this on a different vcr i have, or make an other setup with another vcr and use another vcr/dvr as passthrough, my JVC vcr has more unwanted effects than my Pannasonic or Sony vcr, could it be captured again with a different setup ? that will save some time, in post. btw i use also ProRes422 LT and have no problems with that, others might not agree to that, but judge that for yourself. (better keep it in ProRes before sending it to DigitalFaq, ProRes is a edit codec, the final result can then be rendered in MP4 codec) |
Doing processing on after lossy encoding is really not ideal, but if you have don't have any other option, there are at least some settings that should be different. I don't know what encoder you use, so I'm just going to use general terms. Maybe a better alternative is to ask for help with some avisynth scripts which you can then run on the video yourself.
The test click is encoded as progressive from what I can see, which you don't want. The encoder will be more efficient if it knows the material is interlaced. Additionally the colour data is by default stored at half of both the vertical and horizontal resolution of the brightness7luminance data, so if the encoder thinks the data is progressive, it's going to serously mess up the colour data when resizing. You would probably also want to encode with 4:2:2 chroma subsampling (e.g only halving the colour data in the horizontal direction) to avoid further data loss. The bitrate ought to be way higher, and ideally, use CRF( constant rate factor) mode rather than a set bitrate, this mode will try to keep a set quality level so noisy scenes will get sufficient bitrate. You may also want to consider h.265 here, as it's a bit more efficient than h.264 if you can live with longer encoding time. Quote:
|
Quote:
I have heard this software is not the best..seems this is yet another reason to steer clear of it. |
Quote:
|
Quote:
EDIT: Quote:
|
Agreed with all comments. There are signs of oversharpening, including edge halos, as well as mosquito noise, lack of fine detail and texture, and high-compression detail loss. Overall y-levels exceed 16-235, with some dense blacks and thick shadow with no detail. Chroma is oversaturated when YUV goes to RGB. There is ghosting on the red letters in the background and chroma ghosting on figures during motion when any deinterlacer is used (I used QTGMC and yadif), so encoding interlaced video as progressive didn't help at all. Lossy h264 for restoration and repair isn't really suitable, and the low bitrate makes it even less so.
Overall the low detail, oversharpening, edge noise, and other problems make this look almost like a dim second generation tape dub. Or else the original VHS was in poor shape from the beginning. |
In short.... a new capture might be a good advice, it's not really "handy" to let this done by some one else, since certain things are unknown, and when these tapes are already 2nd generation copies...
a "capture" by (mini)DV device might be even a better a idea :) |
1 Attachment(s)
Thanks for all the feedback so far.
Please check this capture I just did myself - admittedly not a fair comparison to the mp4 above, as it's the original raw ProRes capture file, but I'd like some feedback anyway if any of you guys have a moment to check. Thanks |
Thanks for the sample. But it's not much better, just the same poor quality minus some compression artifacts (but compression noise on motion is still visible. So much for ProRes). Less ghosting but still some edge noise, and really starved of fine detail. Don't you notice the distortion across the top border? It's because the top several pixels of every odd-numbered (bottom) field are warped toward the right. It can't be corrected.
In every respect it's unacceptable. Hope you're not paying real money for it. BTW it looks dim because it's overcast light, but overall it just seems like visually lame output from the player with no dynamic range and poor color density. Perhaps the original recording was made on a really cheap recorder. |
1 Attachment(s)
Quote:
The capture you just saw was played on an FS200 into a Panasonic ES-15 (which was recommended to me) using S-Video. Without the ES-15 or the Datavideo TBC, the playback is so unstable as to be unusable (irrespective of whether we try playing it with a VCR that is equipped with TBC or not) so we need to use something in between. I doubt these devices are causing the issues you specify above, as some very short test captures we did without them (ES-15 or Datavideo TBC) yield more or less identical results. Here is a sample captured as an .avi. Please let me know if you see or can measure a difference between it and the Prores file. Thanks again. |
The source must also be considered, before it was recorded to tape. Don't assume it had no flaws.
- The last AVI recording seems fine, aside from the obvious tearing, and typical VHS noise that need restoration. - The MOV is desaturated and has more blur. The ProRes must be too compressed. You can change it to have less compression. - The MP4 has no tearing, but h264 softening is enabled, and contrast seems high. The AVI without tearing would be most ideal here. |
Quote:
|
Quote:
|
Thank you to all who participated and offered their opinions.
I will proceed with .avi captures for the remaining tapes. Thanks again. |
Site design, images and content © 2002-2024 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2024 Jelsoft Enterprises Ltd.