Should be...

Added left of Pan/Crop - media size

Added right of Pan/Crop - project size

Thanks for that Ed....And by "project size", that carries forward to "render size" too, yes? So for example, if the project is 1080p and I render to 720p -- if the FX is left of pan/crop it'll stay whatever be applied at the media size, right side the render 720p?


I confirmed that with the project set to 1080p, adding Neat to the LEFT of pan/crop, then doing auto-profile, it detects a 4K frame. Adding Neat to the RIGHT of pan/crop then doing auto-profile, it detects a 1080p frame.

Regardless of where it was added, if I change the project setting and then look at the Neat Video settings on a clip, it remains whatever it was at the time it was applied, even if I chose "adjust" as the option. Neat doesn't appear to re-read the project setting once it's applied.

Question: does a noisy 4k clip show as many details as a normal 4k clip? My experience is that noisy 4k clips look even worse than normal 1080 clips.

Thanks for that Ed....And by "project size", that carries forward to "render size" too, yes? So for example, if the project is 1080p and I render to 720p -- if the FX is left of pan/crop it'll stay whatever be applied at the media size, right side the render 720p?

That's a good question, when you choose a lower resolution such as 720p for render the GUI shows Vegas changing the project resolution to 720p, so does that mean Neat video is actually working on your source video at 720p, not 1080p or 4k?

Question: does a noisy 4k clip show as many details as a normal 4k clip? My experience is that noisy 4k clips look even worse than normal 1080 clips.

Any noisy video will lack detail and have a reduced dynamic range making it look worse, with phone cams it looks like they can do a form of synthetic bin'ing of the sensor to reduce noise but it won't be real 4K. At 1080P the camera could use the same area of sensor used for 4K or even the entire sensor (some cameras will crop the sensor for higher resolutions/frame rates) It can gather a lot more information and use that to make a higher quality 1080P.

All cameras work differently, it was common to crop the sensor at 4K to reduce overheating but now it's more common to use more of the sensor and maybe use the same tricks for getting high quality 1080P

Thanks for that Ed....And by "project size", that carries forward to "render size" too, yes? So for example, if the project is 1080p and I render to 720p -- if the FX is left of pan/crop it'll stay whatever be applied at the media size, right side the render 720p?

That's a good question, when you choose a lower resolution such as 720p for render the GUI shows Vegas changing the project resolution to 720p, so does that mean Neat video is actually working on your source video at 720p, not 1080p or 4k?

 

Because the Neat profile *appears* to retain whatever it finds when it first analyzes the frame, my belief is that whether Neat is left OR right, it'll be applying its algorithms based on that original analysis. Which means, to me, it may handle things suboptimal if the render resolution is different from when it was first applied to the event.

SO -- from now on in, I will *always* be applying Neat LEFT of the Pan/Crop before analyzing the frame so it works with source (media) resolution. At render time it won't matter what the target scaling is since the effect is applied pre-scale if it's left of pan/crop.

Question: does a noisy 4k clip show as many details as a normal 4k clip? My experience is that noisy 4k clips look even worse than normal 1080 clips.

The clips I'm currently working on are from Wide-shot 4K footage that sadly didn't have as sharp a focus as I wished, and which we sometimes had to push in 2x . Applying sharpening emphasized the noise, so I applied Neat Video first to the 4K clip, with Temporal sharpening, then added a Vegas sharpen filter. The result? Pretty stunning and clear wide-cam footage!

I've found Neat does a good job with 4K just as it does with 1080, and its own Sharpen-non-noisy-areas does a fantastic job preserving details in my clips.

Thanks for that Ed....And by "project size", that carries forward to "render size" too, yes? So for example, if the project is 1080p and I render to 720p -- if the FX is left of pan/crop it'll stay whatever be applied at the media size, right side the render 720p?

That's a good question, when you choose a lower resolution such as 720p for render the GUI shows Vegas changing the project resolution to 720p, so does that mean Neat video is actually working on your source video at 720p, not 1080p or 4k?

 

Because the Neat profile *appears* to retain whatever it finds when it first analyzes the frame, my belief is that whether Neat is left OR right, it'll be applying its algorithms based on that original analysis. Which means, to me, it may handle things suboptimal if the render resolution is different from when it was first applied to the event.

You could also try Voukoder , when you render it won't change project resolution, if you want a lower render resolution use one of the switches in Voukoder to choose the resolution you like. This will be a slower process. You could then compare the results

Should be...

Added left of Pan/Crop - media size

Added right of Pan/Crop - project size

Thanks for that Ed....And by "project size", that carries forward to "render size" too, yes? So for example, if the project is 1080p and I render to 720p -- if the FX is left of pan/crop it'll stay whatever be applied at the media size, right side the render 720p?


I confirmed that with the project set to 1080p, adding Neat to the LEFT of pan/crop, then doing auto-profile, it detects a 4K frame. Adding Neat to the RIGHT of pan/crop then doing auto-profile, it detects a 1080p frame.

Regardless of where it was added, if I change the project setting and then look at the Neat Video settings on a clip, it remains whatever it was at the time it was applied, even if I chose "adjust" as the option. Neat doesn't appear to re-read the project setting once it's applied.

My assumption:

Left of Pan/Crop will always be Media size

Right of Pan/Crop will always be Project size

Render has nothing to do with anything. If you're rendering to a smaller size, VEGAS will then take the modified Project frame and reduce the size to the render size but the effect would still work at the Project size if right of Pan/Crop.

 

My assumption:

Left of Pan/Crop will always be Media size

Right of Pan/Crop will always be Project size

Render has nothing to do with anything. If you're rendering to a smaller size, VEGAS will then take the modified Project frame and reduce the size to the render size but the effect would still work at the Project size if right of Pan/Crop.

My assumption:

Same as what you wrote EXCEPT I assume at render time, it overrides the project frame size with whatever the render settings are, and that FX applied Right of Pan/Crop will be applied to the new Render size. If so, FX that do some computations (and cache some processing) when they are first built -- tracking, many BorisFX, Neat Video -- may not apply properly if the frame is different from when they cached their data. (thus my NEW decision to always apply those effects LEFT of Pan/Crop)

I don't like "assumptions" ;-) (thus this post)... is @VEGASDerek still on here? Can someone from the Vegas team shed light on exactly what's going on here? (well, don't need the source-code level of "exact" ;-) )

Same as what you wrote EXCEPT I assume at render time, it overrides the project frame size with whatever the render settings are

That's why I told you to try voukoder, it doesn't change project resolution, but you'd also have to have a suitable project that poorly reacts to scaling. The question is do we believe Vegas when it shows project resolution changing to the same as render resolution, is it as if you went into project properties and changed it to render resolution or not.

Help has a nice flowchart for the video signal.

https://help.magix-hub.com/video/vegas/21/en/content/topics/vidsignalflow.htm?skinName=Dark

My assumption:

Left of Pan/Crop will always be Media size

Right of Pan/Crop will always be Project size

Render has nothing to do with anything. If you're rendering to a smaller size, VEGAS will then take the modified Project frame and reduce the size to the render size but the effect would still work at the Project size if right of Pan/Crop.

So -- this rule of thumb has been working STELLAR for me since we had this thread.

I always move Neat Video to the left of Pan/Crop before profiling my clip, and it properly is detecting 4K media on 1080 timelines. AMAZING!

Until today.

Today, in my 1080p59.94 project, when I add a Neat Video effect to my 4K event, and move it to the left of Pan/Crop before analyzing the frame -- it's detecting it as a 1080p (1920x1080) clip instead of 4K.

ARRRRRHHHHHHHHHG!

@RedRob-CandlelightProdctns Maybe you could try using Media FX? Unlike Event FX, Media FX should always use the original resolution of the media.

I agree that the safest and most sensible place to add Neat Video is at the Project Media level, not the track or event level.

Applying Neat Video at the Media level I really don't think is a grand idea... here's why:

1. Neat Video works by analyzing noise in a frame. Because these are stage productions, the lighting conditions vary dramatically, and a great majority of the clips do NOT require noise reduction. Applying generically to the media:
   1. Will needlessly try to reduce non-noisy frames
   2. Will dramatically increase the overall render time
       
2. It didn't matter ;-) I just tried it.
   
   Adding it to the media and clicking "Prepare Noise Profile" gives an information box reading: "You have not selected a frame within the clip before opening Neat Video window. Frames from the beginning of the clip will be used." Clicking OK, then "Prepare Nosie Profile" works, but it still sees it as a 1080p clip (1920x1080), not the 4K media it actually is.

SO -- it seems Neat Video is always taking the resolution from the Project Settings, but the Interlace and possibly framerate settings from the media.

My Media is definitely 4K (Format; MP4 v2-AVC Attributes 3840x2160x32, frame rate 29.97)
Neat Video thinks it is 1920x1080 unless I change the project properties and then it will think it's whatever those properties are set to.

:(

I thought this was worked out... and now it's back to square one!

This isn't really addressed in the "video signal" flowchart posted above (which IS helpful, I don't think it applies here)

Applying Neat Video at the Media level I really don't think is a grand idea... here's why:

1. Neat Video works by analyzing noise in a frame. Because these are stage productions, the lighting conditions vary dramatically, and a great majority of the clips do NOT require noise reduction. Applying generically to the media:
   1. Will needlessly try to reduce non-noisy frames
   2. Will dramatically increase the overall render time

In response to this, you can create a subclip of the trimmed event first, and then apply the media FX to that subclip. This only affects the content of this clip and doesn't disturb other clips.

I don't have a Neat Video though, so I'm not sure how it works...

Applying Neat Video at the Media level I really don't think is a grand idea... here's why:

1. Neat Video works by analyzing noise in a frame. Because these are stage productions, the lighting conditions vary dramatically, and a great majority of the clips do NOT require noise reduction. Applying generically to the media:
   1. Will needlessly try to reduce non-noisy frames
   2. Will dramatically increase the overall render time

In response to this, you can create a subclip of the trimmed event first, and then apply the media FX to that subclip. This only affects the content of this clip and doesn't disturb other clips.

I don't have a Neat Video though, so I'm not sure how it works...

Yip. I sometimes use that technique when I want a particular event *not* to have the media-level FX applied to it.

In this case, how is that any better than applying it to the event itself, left of the Pan/Crop?

It seems when the "Noise Profile" is built for Neat Video, it uses whatever the Preview properties are, and then it saves that. So it's still necessary to change the project properties before having it figure out the noise profile, then changing it back after it's setup. Whether Neat Video is applied to the media-level or left of Pan/Crop has no effect on what it's detecting -- only the project properties.

Applying Neat Video at the Media level I really don't think is a grand idea...

In many tests over the years, I have come to the conclusion that kernel-level filters, including Neat, are almost always better applied to the source media than at a level that has been mediated by any other processing. I've confirmed my visual impressions with various RQM measurement tools.

In your situation, where you say you only want NR on certain scenes, applying it at the Event Level, pre-Pan/Crop and pre-Everything Else, will accomplish the same thing, that of working with your pristine pixels rather than with processed pseudorandom grain. This technique assumes you have not applied any other effects to the Project Media or Track Levels. This is important to me.

Looking farther forward, I have tested AI upscaling, sharpening, and noise reduction at the various levels with both stills and motion video applications, and come to exactly the same conclusion.

Given a recurring desire to apply both a kernel level filter (Unsharp Mask) and AI Upscale at the Media level, I almost always apply the kernel-level filter first.

Applying Neat Video at the Media level I really don't think is a grand idea...

In many tests over the years, I have come to the conclusion that kernel-level filters, including Neat, are almost always better at the media than at a level that has been mediated by processing. I've confirmed my visual impressions with various RQM measurements.

In your situation, where you say you only want NR on certain scenes, applying it at the Event Level, pre-pan/crop and pre-everything else, will accomplish the same thing, that of working with your pristine, rather than processed source.

Looking forward, I have tested AI upscaling, sharpening, and noise reduction at the various levels with both stills and motion video, and come to exactly same conclusion.

Given a recurring desire to apply both a kernel level filter (Unsharp Mask) and AI Upscale at the Media level, I almost always apply the kernel filter first.

What have you found most successful in your AI Upscaling efforts?

I tried the well-self-proclaimed Topaz twice.. both times it was unfathomably slow with only so-so results as compared to other tools, so I never implemented it for our needs.

The hierarchy of processing levels from the bottom up is:

• Project Media Level
• Track Level
• Event Level
• Preview Level, also called the Output Bus.