I've read countless threads on this forum regarding the process of digitizing analog tape-based video. So far I,ve accumulated lots of information but I'm having trouble narrowing it down to a workflow fitting my needs and limitations. Therefore I would be grateful if someone could evaluate my plans regarding what hardware and software to use.
The goal is to shoot a music video for my band on Hi8 and then edit with modern software. The reason for shooting Hi8 in 2019 is purely aesthetical. It has a certain analog look that we miss albeit with a somewhat acceptable resolution. The video is to be released online if everything works out right so there will be a need for deinterlacing.
In short, I want to capture a digital version of the footage as true to the original tape as possible in terms of visual quality. I also want progressive video files that work well in Adobe Premiere for editing.
I have to keep costs at a minimum so buying tons of dedicated hardware won't be possible.
Anyway, here's my gear and planned workflow:
Camera/VCR: Sony Handycam CCD-TRV85 (Records Hi8 XR, XR = 10% increase in "luminance detail" over standard Hi8) (NTSC!)
Capture device: Hauppauge USB-Live2 (Will be using s-video straight from the camera)
Capture software: (I'm planning on using VirtualDub to capture lossless HuffYUV, AVI)
This is where I'm getting sort of lost. Is virtualdub and huffyuv obsolete today? Any recommendation on how to best capture an edit-friendly file format via the Hauppauge? What visual quality can I expect? As said the goal is to keep the analog look of the footage so I'm thinking a lossless capture would be best to avoid any compression messing with the inherent grain of the Hi8-footage(I know from experience that color grading highly compressed video = disaster.) Any recommendations in terms of resolution? Best method for deinterlacing?
Have I got this right or am I completely lost?
Thanks in advance!
Edit: I'm not trying to improve the quality (stabilising, noise removal etc.) I wan't to capture analog video in all its glory with the flaws intact. The main concern is avoiding any digital artifacts when capturing.
Our website is made possible by displaying online advertisements to our visitors. Consider supporting us by disable your adblocker or try DVDFab and copy, convert or make Blu-rays and DVDs! :)
+ Reply to Thread
Results 1 to 10 of 10
Last edited by High On 8; 10th Jan 2019 at 06:48.
HuffYUV is an intraframe codec and thus quite suitable for editing. You shouldn't have any trouble with a reasonably fast processor. Watch out for audio sync problems in VirtualDub; many users, including myself, experience random offsets of captured audio streams. Your biggest challenge will be finding a deinterlacer that you like. Interlaced video never looks the same after it is taken to progressive format because you have to throw away resolution in one form or another.
Consensus is that QTGMC is the best deinterlacer if you have the patience for it. The only advantage of YADIF is that it's much faster. Unfortunately, it performs poorly on diagonal lines in the image.
Avisynth+ x64 QTGMC on fastest preset isn't shabby.
I don't really like slow settings anymore, dislike the blur filter.
I saw a video comparing the different QTGMC-settings and I agree on the blurring not looking very good. I prefer analog video with its flaws, like grain, still visible. I won't try to improve the looks of the capture because in my opinion that can't be done digitally without introducing digital crap to the video. However deinterlacing is a necessary evil. Are there any good guides you know of explaining the basic steps for deinterlacing with QTGMC in Avisynth? I've never written any code so I expect I wouldn't be able to figure it out completely by myself.
Last edited by High On 8; 14th Jan 2019 at 00:26.
High On 8,
I too have a soft spot in my heart for Hi8! I recently digitized almost 60 Hi8's from the 90's and early 2000's and am planning on incorporating a lot of that footage (especially of a radio show I used to do) into a video podcast. Anywho, on the topic of deinterlacing, QTGMC can indeed be good, but for editing in Premiere, I just drop the interlaced AVI clips into a 1080p60 timeline and Premiere converts each field into a frame. (You'll also need to zoom in with the clip's motion settings.) Speeds things up A LOT. It looks great, and as much like an old CRT as possible in the digital world. QTGMC creates detail where there was none in order to smooth out jagged edges. But you don't see those jaggies at regular 60fps (60 fields per second back in the day) anyway. QTGMC is great if you plan to use your Hi8 footage in slow motion clips BUT... In the old days any slow motion or freeze frame from a tape was one field at a time and the jaggies were there anyway. To me, just converting each field to a frame in Premiere is the most authentic look you can get.
In addition to simply dropping interlaced clips into a 1080p60 timeline in Premiere, I also zoom in to crop off the head switching noise at the bottom and slightly ragged edges around the sides and top, even though that results in a touch of detail loss. I like the edge-to-edge coverage, much like an old CRT overscanning.
Here's a finished YouTube video I uploaded recently of a family Thanksgiving in 1998. It was shot on a Sony CCD-TRV65 using the anamorphic 16:9 mode. All I did on it besides letting Premiere do the 1080p60 trick is fix some out-of-range luminance.
Good luck and have fun on that Hi8 project!
There's a difference between preview deinterlacing and render deinterlacing. Only so many algorithms out there and Premiere Media Encoder must be using one of them.
Premiere is weird. If you drop an interlaced 60i clip (59.94) into a 30fps timeline (29.97), it throws away one field and you're left with a half resolution, chunky looking image. But if you put the 60i clip into a 60p timeline, it just doubles up each field and makes a pretty nice, unadulterated 60p video. You would think such a major editing suite would use at least a basic motion detecting deinterlacer like QTGMC for 60i to 30p conversion, but no. I tried Re:Vision Effect's deinterlacer, but for some reason it starts skipping frames on longer clips. Making everything 60fps looks more authentic and smooth anyway. Once i learned that, I stopped caring about deinterlacing so much. EXCEPT... Facebook doesn't allow 60fps yet, so there I just do a field blend out of Premiere and be done with it. Facebook's compression is so awful it hardly matters.
When previewing in the little Program window, Premiere just shows a 30fps video, even if it's a 60fps timeline. Not sure how to get around that.
leeoverstreet that video looks really good! I can't really see any artifacts from the deinterlacing. There's something unique with the way analog tape-based video looks, just like film it has its own character that modern digital cameras don't have. I'm definitely going to try your Premiere method and compare with QTCMT.
Anyway, I just finished my first tests and the results are way better than I expected. I captured some test footage through Virtualdub and a Hauppauge usb-live2 (s-video). Went with Lagarith this time. I tried deinterlacing in Virtualdub using the Yadif preset and the results are definitely acceptable. However, due to the intentionally "hard to deinterlace" nature of the footage some artifacting was visible. I also tried QTGMC after some trouble with getting Avisynth to work. I used the "slower"-preset since the tutorial I followed used that. Also, I had to use Huffyuv since I couldn't figure out how to encode Lagarith in Avisynth. I'm going to try the other presets later. Nevertheless, the result looks very good in my opinion.
Here are the tests, both Yadif and QTGMC. There were some inserted frames during the capture hence some stutter.
Note, video was compressed before uploading to Vimeo.