Universal QTGMC Settings for 10 Years of DV Footage

[taigi]

Hello everyone,

I’m working on deinterlacing and restoring about 10 years’ worth of DV footage using QTGMC, and I’d love some advice on creating universal settings that work across different types of footage. The footage includes three distinct scenarios:

Open-Air Settings (Daytime):
Footage is shot outdoors during the day, under natural light.
Motion is moderate.
Noise is minimal due to the well-lit conditions.

Indoor Club Settings:
Footage is shot in moderately lit indoor clubs, with typical artificial lighting.
Some noise is present, especially in shadowed areas, but not overwhelming.
Motion varies from moderate to fast.

Too Dark Indoor Club Settings:
Footage is shot in very dimly lit indoor clubs with uneven lighting (e.g., stage lights, shadowy areas).
Significant noise and grain are present due to the low-light conditions.

Because I have a large archive of footage, I’m looking for universal QTGMC settings that can handle all three scenarios reasonably well without needing constant adjustments for each clip. I’m willing to consider small tweaks if necessary for specific scenarios.

Here are the settings I’ve been testing so far:

1. Open-Air Daytime Settings:

QTGMC(
Preset="Slow",
TR2=1,
NoiseProcess=1,
NoiseRestore=0.1,
GrainRestore=0.1,
Sharpness=0.4,
SourceMatch=2,
MatchEnhance=0.5,
EZDenoise=0.5,
EZKeepGrain=0.3
)


2. Indoor Club Settings:

QTGMC(
Preset="Slower",
TR2=2,
NoiseProcess=1,
NoiseRestore=0.3,
GrainRestore=0.2,
Sharpness=0.3,
SourceMatch=3,
MatchEnhance=0.7,
EZDenoise=2.0,
EZKeepGrain=0.5
)


3. Too Dark Indoor Club Settings:

QTGMC(
Preset="Slower",
TR2=2,
NoiseProcess=1,
NoiseRestore=0.4,
GrainRestore=0.3,
Sharpness=0.25,
SourceMatch=3,
MatchEnhance=0.8,
EZDenoise=3.5,
EZKeepGrain=0.6
)

My Questions:
Are these settings appropriate for the scenarios I’ve described? Could they be improved?
For fast motion, would reducing TR2 (e.g., to 0 or 1) help preserve motion detail, or will it sacrifice too much noise reduction in darker footage?
Any tips for post-processing after QTGMC (e.g., color grading, encoding)?
Thanks in advance for your advice and recommendations! I’m aiming for a balance between quality and practicality since I’ll be processing a significant volume of footage.

[johnmeyer]

"One size fits all" restoration settings are generally not a good idea. The nature of the problem tends to be different enough from one scenario to the next that not only might one set of settings not work across the different types of source material, but you actually might get artifacts in one case that you wouldn't notice in another. One example is blank blue sky where you can get contouring and other side effects unless you moderate what you do.

Are you using QTGMC for its noise reduction, or do you think you need to deinterlace. Deinterlacing really isn't needed, unless you are resizing or applying temporal noise filters.

What exactly are you trying to get from all this work?

You really need to provide a few samples. It would also help to show us what you get after applying QTGMC. Depending on your source and what is actually wrong with it, the minor improvements may not be worth all the effort.

The most important thing is to make sure you got the DV tapes onto your computer using a 1394/Firewire connector rather than using a capture device. If you used a capture device, you have degraded the video. By contrast, 1394/Firewire provides a lossless transfer to your computer.

[taigi]

The primary reason I’m using QTGMC is for deinterlacing. All the footage was shot using interlaced DV. In addition, I’m also utilizing QTGMC’s noise processing and stabilization features to clean up and restore the footage. Some of the tapes, particularly the indoor ones, suffer from visible noise and grain due to low-light conditions, and QTGMC’s built-in denoising options have been helpful in addressing this. I do understand that one-size-fits-all settings aren’t ideal, but given that I have about 10 years’ worth of footage, I’m looking for a reasonable middle ground that works decently across all scenarios. Fine-tuning each clip individually isn’t practical due to the sheer volume of material I have to process.

Yes, all the footage was transferred to my computer using a FireWire (1394). So, the source files are as clean and true to the original as possible.

I understand the importance of providing samples and will aim to share a few clips maybe soon. The general characteristics of my footage fall into the three categories I described earlier: (1) well-lit outdoor footage, (2) moderately lit indoor footage, and (3) poorly lit indoor footage with significant noise.

My main goal is to make the footage look as clean and watchable as possible while retaining as much detail as I can. I’m not aiming for perfection - just something that looks good enough for viewing on modern displays.

Regarding motion, I’ve noticed that higher TR2 values can sometimes smooth out fast motion too much, which is why I’ve experimented with reducing it for certain clips. Do you think lowering TR2 to 0 or 1 for fast-motion sequences would strike a better balance between noise reduction and motion detail? Or would this create inconsistencies in how the footage is processed?

I’ll also keep an eye on artifacts like contouring, especially in areas like blue skies, as you mentioned. If you have any suggestions for tweaking the EZDenoise, EZKeepGrain, or SourceMatch values to minimize these kinds of issues, I’d love to hear them.

Thanks again for your feedback!

[Sharc]

The primary reason I’m using QTGMC is for deinterlacing.

Why are you deinterlacing at all? Leave it interlaced, and denoise with filters which are dedicated for that purpose.

(Note that some experienced forum members prefer to use QTGMC with settings which does not (or as little as possible) touch the noise, and use dedicated noise filtering after QTGMC.)

Edit: Some interesting thoughts about QTGMC here:
https://forum.videohelp.com/threads/404164-Why-is-QTGMC-so-destructive-and-why-do-so-m...l-recommend-it