Recording, encoding processes for demanding client?

[lordsmurf]

Quote:

Originally Posted by radiokom

In my opinion, in this case he do not want to worry about it.

He's doing what's called "CYA" (cover your ass). I fully understand what and why. It's a pointless exercise to appease TPTB (the powers that be, aka the client). Welcome to organizations. Not just the corporate world, or "corporate America", as they don't have exclusivity to stupid. Plenty to go around.

Quote:

Cheating should be simple. Complex cheating almost always fails, as Remarque said

This is simple. Very cleverly simple. I've done similar tricks to add "detail" to video (grain filters).

Quote:

Originally Posted by SHGMC_2

Going to be honnest, it felt good to see my idea approved by you! My boss also agreed with the strategy. So, we all good on that.

Quote:

Going over Hybrid would be great.
I imagine that you are starting from a huffyuv AVI file. I'll have that ready tomorrow.

Step 1 = drag video file to main window.
Step 2 = adjust/verify settings on video and audio tab.
Step 3 = manually select QTGMC deinterlace settings, manually set field dominance. Consider needs and issues of 59.94 vs 29.97
Step 4= encode

Quote:

I'd also know if there's a good way to force an "interlaced" flag on that ffv1 AVI file just for, you know, pass "the test".

FFV1 usually exists in MKV container, and the container is flagged TFF or BFF (and proper AR).

Quote:

Originally Posted by timtape

The bit length is only a potentiality. The signal chain is only as good as its weakest link, in this case the least noisy link. As with good picture, good audio starts at the very beginning of the production chain and at every step along that chain. The delivery format is only one link at the end of that chain. That is what is so often not understood by people who don't know any better. Usually they're just consumers of other people's work.
Just for interest, VHS can have two soundtracks, one about 9 bit, the other about 14 bit, and that's best case.

Wisdom worth repeating.

Quote:

Originally Posted by radiokom

Hmm.... Something new to me because I never did such foolish thing

I worked at studios. I did a lot of dumbass things because it was required. It was never the studio itself, but rather external labels and and distributors that had asinine requests and requirements. The entire industry is plagued by people who don't know their ass from their elbow, but somehow got into positions of power with limited knowledge of the subject matter.

In other words ... typical to any industry, any company, any org. Nothing is/was special about video.

[radiokom]

OK, I was very curious about that dangerous Bitter and made some tests.
As we know Adobe audition internal is 32bit floating point, so all files are recorded as 32bit not 24.
You can then save them as 24bit.

I took 1 file ripped from CD and converted sample type from 44.1khz16bit to 48khz32bit. Bitter shows 32 bit.
If it is saved as 24bit Bitter still shows 32bit
But not 16.

There are all 3 files - original and converted.

https://files.fm/u/aacqxuzw5v

-- merged --

And here are 2 video fragments. First is with original audio 16bit and second converted to 24bit at export:

https://files.fm/u/7gcdn86tms

So there is a possibility to analyze that .

File Path: D:\TEST\Audio-16.avi
Type: AVI Movie
File Size: 173,5 MB
Image Size: 720 x 480
Frame Rate: 25,00
Source Audio Format: 48000 Hz - 16 bit - Stereo
Project Audio Format: 48000 Hz - 32 bit floating point - Stereo
Total Duration: 00:00:07:20
Pixel Aspect Ratio: 1,0

AVI File details:
Timecode:
User Timecode: 00:00:00:00
Contains 1 video track(s) and 1 audio track(s).
Interleave: 1 : 195.00

Video track 1:
Size is 171.38M bytes (average frame = 907.50K bytes)
There are 195 keyframes.
Frame rate is 25,000 fps
Frame size is 720 x 480
Video format is YUV 422 10-bit CCIR 601

Audio track 1:
Size is 1.42M bytes
Rate is 48000 samples/sec
Sample size is 16 bits

Additional information:
2025-08-20T08:54:08+03:00

File Path: D:\TEST\Audio-24.avi
Type: AVI Movie
File Size: 174,2 MB
Image Size: 720 x 480
Frame Rate: 25,00
Source Audio Format: 48000 Hz - 24 bit - Stereo
Project Audio Format: 48000 Hz - 32 bit floating point - Stereo
Total Duration: 00:00:07:20
Pixel Aspect Ratio: 1,0

AVI File details:
Timecode:
User Timecode: 00:00:00:00
Contains 1 video track(s) and 1 audio track(s).
Interleave: 1 : 195.00

Video track 1:
Size is 171.38M bytes (average frame = 911.25K bytes)
There are 195 keyframes.
Frame rate is 25,000 fps
Frame size is 720 x 480
Video format is YUV 422 10-bit CCIR 601

Audio track 1:
Size is 2.14M bytes
Rate is 48000 samples/sec
Sample size is 24 bits

Additional information:
2025-08-20T08:55:20+03:00

[SHGMC_2]

Won't manually quote everyone on the 24-bit issue... if your software dither absolutely everything the minute there is a non-match in audio specs, then, yes, it'll show some kind of 24-bit...

(See screenshot 1 of post below)

As you can see, there's something CLEARLY wrong here. This is not a normal 24-bit file and you can clearly tell it's an upscaled, dithered 16-bit file.

I understand that just dithering will add noise as seen above but the method I've described for my file will make you see this:
(See screenshot 2 of post below)

As you can see, the 16-bit is entirely hidden by noise much quieter than the VHS noise floor which makes you believe everything is 24-bit. There's no way one can tell it was 16-bit.

How do I achieve a 24-bit file that shows 24-bit? At home, with CD rips (FLAC 16-bit 44.1Khz), using dbPoweramp, I chose Wave then set the number of bits to 24 and leave the rest on "Auto". This will show 16-bit with Bitter even if the file is 24-bit. Why? dbPowerAmp WILL NOT dither or add any kind of noise. You have to tell expressely have to add a Dither DSP. It's the same for Foobar2000 and most encoders anyone could use. Here's an example (sound effect found on freesound.org) with a ZIP file containing the original 16-bit sound and an upscaled 24-bit PCM file:
https://pixeldrain.com/u/Xd8jDjvG

(For copyright reasons, the original file can be found here, free of charge: https://freesound.org/s/155633/ )

As you can see, the PCM file, when played with Bitter, shows 16-bit. Because only padded with 8-bits and no dithering to try to "hide" it. Most converters online will do this. Of course, if you throw that 16-bit file in SoX to upscale to 24-bit with a dither (the reason why you would use SoX to begin with, for great dither), it'll show something similar to the first screenshot above.

Quote:

He's doing what's called "CYA" (cover your ass). I fully understand what and why. It's a pointless exercise to appease TPTB (the powers that be, aka the client). Welcome to organizations. Not just the corporate world, or "corporate America", as they don't have exclusivity to stupid. Plenty to go around.

Yep... and I hate it as much as everyone here that I have to go that route.

Quote:

FFV1 usually exists in MKV container, and the container is flagged TFF or BFF (and proper AR).

Huh. Didn't know that. Yet another "it doesn't make any sense" format decision by them. They want that in a AVI container. GO F'ING FIGURE. MKV would make things so much easier.

Quote:

Step 1 = drag video file to main window.
Step 2 = adjust/verify settings on video and audio tab.
Step 3 = manually select QTGMC deinterlace settings, manually set field dominance. Consider needs and issues of 59.94 vs 29.97
Step 4= encode

Questions:
- What deinterlace preset do you use? Fast (default)? Slow?
- What % of "Analyze" de you use? Default is 5%.
- The framerate is something that I knew about. I know that, technically, when playing back interlaced content, you play at half the framerate. So, it's important to record at 59.94. However my TV Wonder 600... if I try to record at that framerate, VDub will insert a frame for every frame (duplicating each frame basically). Unless this is something that needs to happen, I can only record at 29.97 to minimalize inserted frames. So what do in this case?

For my case, I went with 5%, slow and the original 29.97fps from the huffyuv.

Quote:

I worked at studios. I did a lot of dumbass things because it was required. It was never the studio itself, but rather external labels and and distributors that had asinine requests and requirements. The entire industry is plagued by people who don't know their ass from their elbow, but somehow got into positions of power with limited knowledge of the subject matter.

In other words ... typical to any industry, any company, any org. Nothing is/was special about video.

Yep! I see it often as well. Just recently, I stumbled upon a clear upscale from low quality lossy file (just by looking at spectograms, it's extremely obvious) that's maybe equivalent of MP3 VBR V4. It was encoded at 24-bit 48khz. The mind-boggling thing about it as it shows as real 24-bit with Bitter (like second screenshot above). Sorry, your fucking master sucks. I understand that you might have lost the original master and want to put something on sale but, when this is the case, it's 16-bit 44.1khz. You are already embellishing a turd. You don't need to give it a royal treatment.

[Aya_Rei]

Quote:

Originally Posted by SHGMC_2

Questions:
- What deinterlace preset do you use? Fast (default)? Slow?


- The framerate is something that I knew about. I know that, technically, when playing back interlaced content, you play at half the framerate. So, it's important to record at 59.94. However my TV Wonder 600... if I try to record at that framerate, VDub will insert a frame for every frame (duplicating each frame basically). Unless this is something that needs to happen, I can only record at 29.97 to minimalize inserted frames. So what do in this case?

For my case, I went with 5%, slow and the original 29.97fps from the huffyuv.

I believe LS suggests the Faster preset, I myself use slower.

You do not record your capture at 59.94 fields per second. Analog tapes are supposed to be captured at 29.97 interlaced fields per second (25 fields per second for PAL) and then de-interlaced to 59.94 progressive frames per second. (Or 50 FPS if the source is PAL)

Or de-telecined back to 24 FPS if the footage was originally shot with that target FPS (IE motion pictures)

[radiokom]

If it is constant noise it can be recognized under spectral display. By adding 24 bit white noise you can fool Bitter, but not spectral display. Of course all depends how deep your client can analyze that. A good forensics lab would decipher whatever you do

[SHGMC_2]

Quote:

Originally Posted by Aya_Rei

I believe LS suggests the Faster preset, I myself use slower.

You do not record your capture at 59.94 fields per second. Analog tapes are supposed to be captured at 29.97 interlaced fields per second (25 fields per second for PAL) and then de-interlaced to 59.94 progressive frames per second. (Or 50 FPS if the source is PAL)

Or de-telecined back to 24 FPS if the footage was originally shot with that target FPS (IE motion pictures)

I looked at my "slower" result after a 2-pass encode and it looks great imo.

Can you teach me how to do that in Hybrid? Everything we do is NTSC. We've never digitized PAL.

Quote:

Originally Posted by radiokom

If it is constant noise it can be recognized under spectral display. By adding 24 bit white noise you can fool Bitter, but not spectral display. Of course all depends how deep your client can analyze that. A good forensics lab would decipher whatever you do

Heh. Sure. But I can absolutely make sure you don't see it in the high frequencies with a single effect.
But then, your regular user will most likely deduce it's just the VHS noise looking at the specs. And who's going to a forensic lab? If you want to spend several thousand to spot a fake, be my guess. No one's going to go that far.

[radiokom]

I wonder how difficult it would be to modify VirtualDub to record audio in 24 bit? 24bit has been the standard for over 20 years now and HDD space is really not a problem. I believe your case is only one of the first, there will be more.

[Aya_Rei]

Quote:

Originally Posted by SHGMC_2

I looked at my "slower" result after a 2-pass encode and it looks great imo.

Can you teach me how to do that in Hybrid? Everything we do is NTSC. We've never digitized PAL.

Oh, you're probably de-interlacing them with the final frame rate still being 29.97

For 59.94, click on the checkbox that is labeled 'bob'

For de-telecine, there is a dropdown box on the right, click it and select IVTC. Leaving IVTC's settings at default it fine enough for me.

[SHGMC_2]

Quote:

Originally Posted by Aya_Rei

Oh, you're probably de-interlacing them with the final frame rate still being 29.97

For 59.94, click on the checkbox that is labeled 'bob'

For de-telecine, there is a dropdown box on the right, click it and select IVTC. Leaving IVTC's settings at default it fine enough for me.

Yeah, that's what I did. The result still looks really good to my eyes (for what's a VHS is worth anyway).

Two things with what you told me:
- In the dropdown (which I assume is the "Auto Deinterlace Handling" one), there's "VIVTC" and "TIVTC". Not just "IVTC".
- The "Bob" checkbox disappear when selecting the other thing from the drop-down. It's only there if the de-interlacer is just QTGMC. There's no such checkbox with either of the above.

Using version 2025.07.27.01

Also, my huffyuv's been seen as bottom-field first (don't remember if this was told to me - there had been so much going on here) in Hybrid. I thought it was top-field first. I guess that's correct and I've used BFF for my previous encode.

Also, just to make sure, the client expect their MP4 version to be de-interlaced but they don't say how to do it or how if they expect the de-interlacing video to be 59.94. They just say "Resolution: same as preservation (match original)". I just don't know if it's worth the trouble at this point.

EDIT: To make sure, the original huffyuv is seen as 29.97 - just in case that was not clear.

Quote:

I wonder how difficult it would be to modify VirtualDub to record audio in 24 bit? 24bit has been the standard for over 20 years now and HDD space is really not a problem. I believe your case is only one of the first, there will be more.

I don't think it would be too difficult for a good programmer. Detect what the selected capture device can do (Windows API can tell you that I'm pretty sure, since you can set that yourself for your sound card, speakers, etc.) and show the meaningful options.

As for the client, they exist since... 1856. I'm not joking. So, since they started digitalizing their archives, I wouldn't be surprised if they've been asking for 24-bit audio for analog video for a very long time now. So, I'm definitely not the first.

[Aya_Rei]

That's the point, just saying that TIVTC is used if you want to de telecine footage, QTGMC for de interlaceing.

Yeah Hybrid auto flags the huffyuv .avis as bff since those files have no flags set, you need to manually override it to top field first.

What's this about resolution? Needs to be the same as the raw file (720x480) I mean, it should be downscaled to 640x480 since that is a 4:3 resolution, no need to mess with pixel aspect ratio (PAR), you can keep that at 1:1 with a 4:3 resolution

[SHGMC_2]

Quote:

Originally Posted by Aya_Rei

That's the point, just saying that TIVTC is used if you want to de telecine footage, QTGMC for de interlaceing.

Yeah Hybrid auto flags the huffyuv .avis as bff since those files have no flags set, you need to manually override it to top field first.

What's this about resolution? Needs to be the same as the raw file (720x480) I mean, it should be downscaled to 640x480 since that is a 4:3 resolution, no need to mess with pixel aspect ratio (PAR), you can keep that at 1:1 with a 4:3 resolution

Right. I wasn't going to go that far tbh. They'll see a progressive video that look different from the Master, assumes it's deinterlaced properly and move on. I mean, with all these crazy specs... it's clear they don't know what they're talking about anyway. If my client had absolute knowledge in this (example: LS) and asked me to do this for them, I'd go the extra mile and do what you said. This is not worth it. I'll only correct the BFF mistake with a final re-encode. This rip will be sent in the week of Sept. 1st to see what they say.

I'd love to set the PAR to be 4:3 (I think it's 8:9 ratio?) and they did say to match original resolution. So, if they see the preservation master being 720x480 but the MP4 being 640x480... again... I'll remind you they ask a H(x)264 16mbps MP4. It's another CYA situation...

[Aya_Rei]

Well if it needs to be 720x480 you can make the output par be 8:9, this might be helpful
,
I tend to upscale my footage to 1920x1400 for YouTube with an output PAR of 1:1 since the resolution is already 4:3

[lordsmurf]

Quote:

Originally Posted by SHGMC

Questions:
- What deinterlace preset do you use? Fast (default)? Slow?
- What % of "Analyze" de you use? Default is 5%.
- The framerate is something that I knew about. I know that, technically, when playing back interlaced content, you play at half the framerate. So, it's important to record at 59.94. However my TV Wonder 600... if I try to record at that framerate, VDub will insert a frame for every frame (duplicating each frame basically). Unless this is something that needs to happen, I can only record at 29.97 to minimalize inserted frames. So what do in this case?

- faster --- unless obvious issues, then Slow
- Analyze?
- "bob" adds 59.94 to Hybrid' QTGMC
- the most important is to overwrite(override) field dominance to TFF away from BFF default. (I need to mention "overwrite vs. override" to selur.)

Quote:

Originally Posted by Aya_Rei

You do not record your capture at 59.94 fields per second. Analog tapes are supposed to be captured at 29.97 interlaced fields per second (25 fields per second for PAL) and then de-interlaced to 59.94 progressive frames per second. (Or 50 FPS if the source is PAL)

Actually...
- 59.94/50 fields
- 29.97/25 interlaced frames

Quote:

Originally Posted by SHGMC_2

If you want to spend several thousand to spot a fake, be my guess. No one's going to go that far.

Never going to happen.

Quote:

Originally Posted by SHGMC_2

(for what's a VHS is worth anyway)

Never say that. VHS is far better than people give it credit for. Bad players, bad software, bad settings, is what leads to bad VHS video. It's not intrinsically terrible, but it has limitations and ceilings. The thread is for a client that doesn't understand the upper end, but don't denigrate it to the lower end.

Quote:

Also, just to make sure, the client expect their MP4 version to be de-interlaced but they don't say how to do it or how if they expect the de-interlacing video to be 59.94. They just say "Resolution: same as preservation (match original)". I just don't know if it's worth the trouble at this point.

Given the client, just do 59.94.

Quote:

As for the client, they exist since... 1856. I'm not joking. So, since they started digitalizing their archives, I wouldn't be surprised if they've been asking for 24-bit audio for analog video for a very long time now. So, I'm definitely not the first.

And I bet they're been fired as a client before.

Quote:

Originally Posted by Aya_Rei

That's the point, just saying that TIVTC is used if you want to de telecine footage, QTGMC for de interlaceing.

Telecine isn't always smoooth, or non-mixed, so deinterlace is the lazy way. For this client, I'd be lazy here. It's already taking too much time, and the VHS source is likely mixed, even if some exists as telecined. I would not mess with it.

Quote:

Originally Posted by Aya_Rei

What's this about resolution? Needs to be the same as the raw file (720x480) I mean, it should be downscaled to 640x480 since that is a 4:3 resolution, no need to mess with pixel aspect ratio (PAR), you can keep that at 1:1 with a 4:3 resolution

Yep.

Quote:

Originally Posted by SHGMC_2

they ask a H(x)264 16mbps MP4. It's another CYA situation...

LOL!

Bloat for sake of bloat. Good gosh.

Quote:

Originally Posted by Aya_Rei

Well if it needs to be 720x480 you can make the output par be 8:9, this might be helpful
I tend to upscale my footage to 1920x1400 for YouTube with an output PAR of 1:1 since the resolution is already 4:3

This client is the sort of person that will stretch VHS to 16x9, and think it's great.

[SHGMC_2]

Quote:

- faster --- unless obvious issues, then Slow
- Analyze?
- "bob" adds 59.94 to Hybrid' QTGMC
- the most important is to overwrite(override) field dominance to TFF away from BFF default. (I need to mention "overwrite vs. override" to selur.)

For Analyze, see screenshot atatched. I circled the relevant field.

Quote:

Given the client, just do 59.94.

Started the 5th re-encode...

Quote:

and the VHS source is likely mixed, even if some exists as telecined. I would not mess with it.

Looking at the content, I'm at least 90% sure that it's not mixed. It was a camcorder recording. And, yeah, telecine is something I don't want to mess with, tbh.

---

Hybrid has this thing where it forces a resize if you want to change the PAR. I didn't see it the first time and it was like 720x426. I was like "wtf". Now, I forced it back to 720x480 but I think, for future transfers, I am thinking of using this ffmpeg line:

ffmpeg -i input.mp4 -c copy -aspect 4:3 output.mp4

This sets the PAR to 4:3 without re-encoding anything. This will even work on AVI files (forcing to have a PAR flag even it shouldn't have).

I'd like to have opinion on the command above.

[radiokom]

You need to set upper (or top) field first.

[aramkolt]

I never really paid too much attention to the bit depth of the audio, but it turns out I've been capturing 24 bit audio all along with the AJA KiPro when I just checked the specs. Per the manual, it captures 24 bit audio from both embedded SDI and the analog inputs. The downside is that it captures the audio channels as separate audio "tracks", but you can use an FFmpeg script to merge the two Mono tracks into one stereo track without re-encoding the video afterwords. Sounds like from there, it'd just be converting to this V210 format and you'd be golden.

KiPro only accepts component video or SDI, so you'd either need a TBC that outputs one of those, or to run your pre-timebase corrected composite or S-Video into it through something like a blackmagic analog to SDI mini converter (which can also embed analog audio into the SDI stream at 24bits). It doesn't drop frames as long as the signal is pre-timebase corrected.

[SHGMC_2]

My final encode with Hybrid turned out great. 720x480, 4:3, de-interlaced @ 59.94...

I kept it at "slow" because I was really surprised how good it looked like that.

Quote:

Originally Posted by radiokom

You need to set upper (or top) field first.

It is set to TFF in Hybrid. I can see it in my screenshots. Unless you are ferring to something else.

Quote:

Originally Posted by aramkolt

I never really paid too much attention to the bit depth of the audio, but it turns out I've been capturing 24 bit audio all along with the AJA KiPro when I just checked the specs. Per the manual, it captures 24 bit audio from both embedded SDI and the analog inputs. The downside is that it captures the audio channels as separate audio "tracks", but you can use an FFmpeg script to merge the two Mono tracks into one stereo track without re-encoding the video afterwords. Sounds like from there, it'd just be converting to this V210 format and you'd be golden.

KiPro only accepts component video or SDI, so you'd either need a TBC that outputs one of those, or to run your pre-timebase corrected composite or S-Video into it through something like a blackmagic analog to SDI mini converter (which can also embed analog audio into the SDI stream at 24bits). It doesn't drop frames as long as the signal is pre-timebase corrected.

We're not buying another solution for this. They know what our hardware is. If that was not OK with them, then, they shouldn't have given us VHS tapes.

[radiokom]

Quote:

Originally Posted by SHGMC_2

It is set to TFF in Hybrid. I can see it in my screenshots. Unless you are ferring to something else.

I am confused by settings: input scan type "bottom field first" and checked "overwrite input scan type to top field first". I am not familiar with hybrid, but if your video has interlace flags in my opinion it should be visible as "top field first" what "bottom" is doing here?

[SHGMC_2]

I would like to take the opportunity to say Thank You to everyone that has stopped by this thread to give their advice. I really appreciate it and I feel I learned a lot!

Quote:

Originally Posted by radiokom

I am confused by settings: input scan type "bottom field first" and checked "overwrite input scan type to top field first". I am not familiar with hybrid, but if your video has interlace flags in my opinion it should be visible as "top field first" what "bottom" is doing here?

I think LS mentionned it earlier. Hybrid misdetects avi Huffyuv interlaced footage from most analog video as BFF. That is why it needs to be overwritten to UFF.

[radiokom]

Quote:

Originally Posted by SHGMC_2

I think LS mentionned it earlier. Hybrid misdetects avi Huffyuv interlaced footage from most analog video as BFF. That is why it needs to be overwritten to UFF.

OK, Thanks!