Great comparison video as usual! Using DV as a starting source is interesting because you might find that something that was analog to begin with may be "more improved" by various deinterlacing algorithms. This is presuming the original footage was shot on DV as opposed to being converted to DV from VHS by something like the ADVC110. Hardware deinterlacers might be an interesting thing to compare for future videos as well. The downside to hardware deinterlacers though is that you can't go back to the "source" material if a better deinterlacing method comes along later of course.
Yeah, I agree that natively shot MiniDV footage is different from Video8 footage captured with even a good SD capture device. Also I recorded the DV footage outside with plenty of cloud diffused light, which was as good as MiniDV is going to get. Anyway I assume (hope?) the results carry through to SD footage.
I kind of wonder about this 59.9 fps gospel for 29.9 video. Let's say your capture chain actually delivers ~30 half-frames per second. Now, to deinterlace, you're either going Bob (double the scan line in each half-frame to fill the screen and cut your vertical resolution) or Weave (comine/blend the 2 half fields, keeping your vertical resolution but getting combing artifacts and lower temporal resolution). Or a combination of both (bwdiff). But in neither case are you going to create "more" frames than 30. (Yes, I hear some advanced algorithms look across many frames/field and interpolate entirely new frames, these could output more than 30 fps I guess, but it is unclear to me if they are allowed to do so in the processing chains you and I use.) If you crank the frame rate up AFTER the deinterlacer, well, yes, that's maybe going to do temporal interpolation to create new frames, and the increased, synthesized temporal resolution may make motion look better to the eye, but that would be true if that was done to any lower FPS source material. I just did a test with some fast motion video of a pumpkin rolling down a hill (don't ask) and did the scale/deinterlace/frame-rate transform with this Perplexity-generated ffmpeg command: ffmpeg -i input.mp4 -vf "bwdif=mode=1:parity=tff:deint=all,scale=1440:1080:flags=lanczos" -r 59.94 -c:v prores_ks -profile:v 3 -vendor apl0 -bits_per_mb 8000 -pix_fmt yuv422p10le -c:a pcm_s16le output.mov Looking at the results in Final Cut with all of its local viewer deinterlacing and field-forcing turned off what I see is a weird repeat of earlier frames, seemingly to "pad" the time. So when the camera is zooming in, looking at it frame by frame, I see it go in-out-in-out-in-out. Mind you, it looks fine running "at speed." But doing a simple upscale without frame rate change the zoom just goes in-in-in-in as you'd expect. And to my eye looks no different, and maybe better than the 60 fps version. I get that, intuitively, giving the deinterlacer double the frame rate room to "spread out" seems logical, but I don't think the deinterlace filters we commonly use do that, or, if we use an exotic deinterlacer like VapourSynth that might want to create new frames (I don't know if it actually does) if it would be allowed to do so since framerate conversion and deinterlacing are separate, isolated blocks in our processing chain. Right now, for me, I don't see enough (or any) difference between 29.97 and 59.94 to justify the massive increase in the size of my Prores 422 intermediate files. But I'm interested to know where my thinking has gone wrong since so many people seem to swear by this 60 fps mantra.
I read your comment a few times as there is a lot to it. I'm not an expert in deinterlacing or the theories. As a next step I would suggest you repeat the steps but this time add a line of code to indicate if it's a top field first or bottom field first source video (ie, SD capture or DV transfer). The symptom you described with the pumpkin roll reminds me of what happens when TFF is used on a MiniDV video, and vice versa. Then I suggest you try QTGMC deinterlacer to verify if it does the same thing or not in the pumpkin video.
It starts with the actual fps of your source. Interlaced NTSC has both 29.97 and 30.00, but broadcast is 29.97 as it was easier to transmit than 30.00. So when you deinterlace, you're joining both top field and bottom field into a single frame, so 29.97 becomes 59.94. Increasing it to 30/60 is possible, but it speeds up the audio. But as for combing artifacts, it comes down to the algorithm you choose... the better the algorithm, the less chance of artifacts.
@@michaelbinbc No, I don’t think that’s true. That’s not how deinterlacing algorithms work as I tried (poorly) to explain in the comment above. I’m traveling now so can’t get into it further, but I may return to this when I return home.
@@WardCogo ask on an actual video editing site like Blackmagic... the makers of Davinci Resolve editing software. Industry professionals will give the same response. Bob only uses 1 field at a time, but interpolates the missing data, while weaving the 2 fields into a single frame at 2x the fps gives all the data and looks far superior. And broadcast interkaced NTSC in North America is 29.97, not 30. They have 30fps videos, but not for broadcast. It comes down to data rate and the ability to broadcast both color and stereo audio. But hey... believe what you want I guess.🙄
Do you use anything to denoise? I have found the vhs tapes when digitized often have a fair amount of audio hiss to them. I've messed with premiere's denoise before and it does a decent job.
Hybrid has a number of denoise tools but I'm not an expert. I have used the denoise option in QTGMC. Also Temporal Denoise. I have seen other people refer to other ones in their work. I err in the side of minor denoise so as not to do damage to the video. I'm hoping tools in the future will be able to look at your raw video and apply the best denoise (and other) filters the way that video restore experts currently do thanks to their vast experience. Another good option today is the Neat filter plugin. It works in Resolve, Premiere and even VirtualDub. It gives good results and I have seen videos where its sharpening tool works well in combination with its denoise tool. Regarding audio hiss, Adobe Audition can deal with that.
Hello all, I'm new to this and was wondering, is there a video or tutorial to help me get started with deinterlacing some Video 8 PAL analogue footage to digital with my MacBook Pro? I have OBS and CapCut but bewildered with so many settings I get confused. Would really appreciate a step by step video or tutorial. Hope someone can help. Many thanks. Steven. U.K.
@@videocaptureguide No not converted yet, too many setting to know what to choose. I have OBS and capcut. I understand Hybrid is quiet costly. This is mainly needed for a one off conversion for now at least. I'm using my MacBook Pro M1 Max, not WIndows system for info. Thanking you in advance. Steven.
I will publish a future video with more details on how to deinterlace. Hybrid is a free software. And while it can feel overwhelming, there are only a few screens that one needs to use.
You didnt deinterlaced in Pr. It was just resized and framerate was poorly streched. To actually get 2x framerate you have to use deinterlacer in flow. Something like Yadifx2 which used both fields as separete frames instead of trying to combine them.
There is no need to de interlace when using Premiere Pro with a 60 FPS sequence. The 59.94 fields are easily interpolated into 59.94 frames canceling out the interlaced fields. Motion will look better at 60P as opposed to 30P. Having said that I don't see how Video Capture Guide's DV camcorder had such horrible image quality. We should have seen what the video looked like at it's native resolution as a guide for image quality. For this video, Video Capture Guide should have used Media Encoder as opposed to using Premiere Pro because you cannot do over the shoulder editing using Hybrid but you can with Premiere Pro. The link below might be worth watching. ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-vqCi9B_7mdU.html
@TechTVusa I would be willing to send you a WeTransfer with the raw DV file taken with the TRV-17 so that you can to deinterlace in Premiere Pro 2024 to see if you get the same results as did I. The point of the comparison video wasn't to show off how good or bad the original TRV-17 camcorder is, but to compare the results of Premiere Pro versus QTGMC-in going from interlaced 29.97fps to progressive 59.94fps. Regarding Adobe Media Encoder, I just exported it the video from my Premiere Timeline via Media Encoder and looked at the two videos side by side and they are identical. If there is a specific setting you want me to try in Media Encoder, please let me know. But the settings appear to mirror the ones in Premiere Pro. Perhaps you can do a comparison video using your own DV footage and try Premiere Pro versus Hybrid (QTGMC) and then post the findings on your channel.
@@videocaptureguide I appreciate your response and willingness to collaborate. My problem with the video is you should have used Media Encoder as opposed to Premiere Pro for this video. My video will demonstrate why. I will give you a shout out in my video response (you might pick up some subs) and actually revolutionize your thought process at the same time. My thought process has been altered as well. Having said that my video will be much different than anything you or I have done thus far.
@@TechTVusa I mean it didnt deinterlace it automatically. It just burn interlaced video into progressive and will mess it up with compression. Basic mistake.
@@stanislavnepochatov8381 I never said premiere pro deinterlaced the video. Please reread my post. It is interpolating the video from interlaced fields to progressive frames eliminating interlacing artifacts.. That it did do. Look at my video quality (link below). Look at how well Premiere Pro can play back 480i and 1080i video in realtime in the year 2024. Keep in mind you could do the same thing using FCPX, iMovie and even the Edius software. Keep in mind the Edius software can use the Lanczos method for scaling. That being said I thought this video was going to be the most amazing VHS/Hi 8 video ever at the very beginning but instead it turned into a horrible video with very bad commentary. Keep in mind Premiere Pro is a video editor not transcoding software. Media Encoder is transcoding software. Wait until my video posts and I highly doubt you will disagree with any of my comments. ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-vqCi9B_7mdU.html
Before I make a response video I should ask. Are you going to make a Premiere Pro workflow video like other people have asked for? I am asking because if you plan on making that video soon I may not need to make a response video.
I have not yet been able to successfully install an older version of Pr on my other Windows 7 computer. I keep getting an error. So I can't capture DV with Pr. Is that what you are referring to? Is it your view that transferring DV via Pr instead of WinDV will somehow improve the eventual deinterlacing of the video inside Pr?
@@videocaptureguide WinDV has options. That is a big NO NO! Edius, FCP and Premiere Pro do not have options for rercompression or any other options for that matter. Why would they? Having said that it is the ease of use and what can be done when using Premiere Pro, Media Encoder and Encore. The Holy Trinity is hard to beat. As I stated in other posts the software is the most important part of the equation. I admit Encore is now defunct as is video video capture in Premiere Pro but Premiere Pro still does the job very efficient. The scaling that was performed in this video was not something I would do for a multitude of reason. How are you adding tiles, color correction and cleaning up the audio? That is what I meant by an efficient work flow. The Holy Trinity was hard to beat if you had several clients per week giving you ten tapes each. The other software you used was like working a combination lock in order to get the settings correct. There are few things I found odd about the video. It was not what I was expecting.
“WinDV has options. That is a big NO NO! Edius, FCP and Premiere Pro do not have options for recompression or any other options for that matter.” The only option in WinDV type 1 or type 2 DV. The internet says the only difference is that type 2 contains a second copy of the audio track explicitly marked as audio in the AVI container. And there is no recompression setting in WinDV. So, I think WinDV is a big YES YES. But I am open to trying Premiere Pro to transfer, but I haven’t been able to find/install an older copy. I would very much like to do this comparison to either confirm what you are saying or confirm what everyone else is saying. Because everything else I have read states it makes no difference which application you use to transfer DV footage from a camcorder to a computer via firewire. It’s just data. “The scaling that was performed in this video was not something I would do for a multitude of reasons.” Can you clarify what you mean? Do you mean that you don’t recommend scaling any video because you’d rather keep it at its native resolution and let your TV upscale it? Or do you mean that the method I used in Premiere Pro is not the best one, and that there is a better way with different settings? If so, please let me know so I can redo it. “How are you adding tiles, color correction and cleaning up the audio? That is what I meant by an efficient work flow. The Holy Trinity was hard to beat if you had several clients per week giving you ten tapes each. The other software you used was like working a combination lock in order to get the settings correct.” I agree that one software that does things good or great, is probably a good choice. However, I think deinterlacing is important and has a big impact on the final visual. I think taking the time to go from 480i 29.97fps to 480p 59.94 fps with the best tool (QTGMC) is worth it.
@@videocaptureguide This video did have some good information in it for QTGMC but at the same time it was horrific to watch. Was your goal to provide helpful information or was your goal to simply show everyone you found a way to get better image quality using the QTGMC software as opposed to using Premiere Pro? I should state right now. You didn't necessarily achieve better image quality using QTGMC and you forgot to mention the side effect with using your method. That is why watching this video was like watching a five year old play with crayons. Don't get me wrong. I think there could be a time to use QTGMC but this video did not demonstrate that. You need to provide a little more information and a little better paradigm. I think you now realize there are some serious problems with this video. As I stated some people were expecting a Premiere Pro workflow video not a transcoding video. I think you are starting to see the huge flaws in your video. Using Premiere Pro was a huge flaw and a complete waste of time for this video. You should have used Media Encoder for this video. Having said that I will create a video that show the power of the Holy Trinity Vs the cheap software you use. I will end by saying at least you are testing things out for yourself but going forward into the future I hope you will think things out a little better. Keep in mind I even offered to do a video chat with you.
Very informative video. Thank you for the follow up. Two questions: 1. why not use the deinterlace option in Premiere under field options; does it cause issues? 2. Is virtualdub good for deinterlacing after capturing? Thanks again for your time and research.
VirtualDub's deinterlacing filter only gives you a few options. It's VERY limited. Same with Handbrake. Despite its humanity-hating user interface, Hybrid and QTGMC produce far, far superior results.