[Mads Nybo jørgensen]

Hey Steve,

Admittedly, I asked Gemini, as I was curious – and too busy to go to the source:

The author’s intuition in the second half is spot on, though the math in the first part has a minor technical discrepancy. Premiere Pro will not detect or strip out duplicate frames on its own.

1. The Math Check (Premise 1)

• Duration & Count: Slowing down to 10% speed expands the clip’s duration by a factor of 10. If a 1-second 24 fps clip (24 frames) becomes 10 seconds (240 frames), Premiere needs 10 frames for every original frame. That means it inserts 9 synthetic frames between each pair of originals (1 original + 9 generated = 10 total), not 10 inserted.

• Effective Cadence: The reasoning on perceived time is conceptually sound: each frame advances the captured action by 1/240th or 1/300th of a second, producing continuous, fluid slow motion provided the algorithm doesn’t artifact.

2. What Happens with Baked Duplicates (Premise 2)

Premiere Pro treats any imported video file as raw raster data. It has no automated cadence detection to realize that a baked file has repeated frames ($A, A, B, B$). It strictly evaluates sequential frames as they exist in that source file.

If a clip with baked duplicate frames is stretched with Optical Flow, here is what the engine calculates:

• Between Frame 1 ($A$) and Frame 2 ($A$):

  The pixel values between the two frames are identical. The motion vector field calculates zero velocity ($\Delta x = 0, \Delta y = 0$). Optical Flow synthesizes intermediate frames that are exact clones of $A$. The action completely freezes.

• Between Frame 2 ($A$) and Frame 3 ($B$):

  The engine suddenly detects a large displacement because all the movement from two frame intervals has occurred across a single frame step. It calculates vectors from $A$ to $B$ and interpolates motion across this window.

• Between Frame 3 ($B$) and Frame 4 ($B$):

  Zero motion again. It generates identical clones of $B$.

The Resulting Artifacts

Feeding baked duplicates into Optical Flow causes two distinct problems:

• Pulsing / Hitching Motion: Instead of steady slow motion, the clip will play with an uneven, stuttering cadence: freeze → morph → freeze → morph.

• Severe Warping and Tearing: Because the motion between $A$ and $B$ is twice as large as it would be in native footage, the displacement often exceeds the algorithm’s vector search window. This causes tearing, rubbery distortion, and boundary smearing.

How to Fix It Before Applying Optical Flow

If working with a file that already has baked duplicated frames (e.g., a 24 fps file doubled to 48 fps with duplicate frames, or a 25 fps file padded to 50 fps):

1. Interpret Footage: If the file’s container frame rate was doubled to accommodate the copies, right-click the clip in the Project Panel → Modify > Interpret Footage, and conform it back to the true base capture rate.

2. Remove the Duplicates on a Sequence: If the file has duplicates at standard timeline frame rates, drop it into a sequence matching its real capture rate, set speed to 200% with Frame Sampling (which drops every second frame), export a clean master, and then bring that master back in to apply 10% speed with Optical Flow.

Hope this helps?

It would be interesting to compare this with FCPX, Davinci and AVID + RevisionFX Twixtor, for how the different “optical flow” or similar processes works.

Atb
Mads

[Mads Nybo jørgensen]

Hey Shin,

Many NLE editors now comes with excellent noise reduction baked in.

If I have to go further, I use Izotope.
Although I have the bigger suite, I originally got the starter plug-in pack, and never looked back:
https://www.izotope.com/products/rx-elements
(Izotope is now owned by BorisFX)

Obviously, the spanner in the works is A.I..
As in that many of the big A.I. engines out there can either clean up the track, or even vibe code the effect as a plugin for you.

One thing to look out for, is that in the early days of Adobe PPro “new” noise reduction filter, it at times felt that Adobe transcribed the clip, then “learned” the voice characteristics, then in the back-ground, “re-recorded” the voice to the trascribed clip an d repaced the old noisy voice-track with that “clean” recording.
This you might imagine, could give some hillarious results, involving phantom words in the middle of a sentence.
Adobe is getting better, but no matter what you use for processing, you should still ensure that the spoken word remains the same…

Hope this helps?

Atb
Mads

izotope.com

Professionally repair and enhance audio with RX 12 Elements

Fix audio flaws in seconds with RX 12 Elements, an intelligent set of 6 repair plugins designed to effortlessly tackle common audio issues.

[Mads Nybo jørgensen]

Sorry, if the above looks garbled – blame it on WordPress…

However, here is a 7-day link to what Gemini told me:
https://share.google/aimode/xMLnlBrSKAdY0awnI

Atb
Mads

share.google

Google Search

Google Search

[Mads Nybo jørgensen]

Hey Tim,

Did a search, using my new friend Gemini, but centred it on Black Magic Fusion.

I have no idea as to whether this will work:

To export multi-layer EXR files in Blackmagic Fusion, you need to combine your different passes (such as Beauty, Z-Depth, Ambient Occlusion, or Motion Vectors) into a single image stream using Channel Booleans or Merge nodes before routing them into a single Saver node.

Here is how to set up and export a multi-layer EXR file.

<b role=”heading” data-sfc-cp=”” jsaction=”” jscontroller=”gjwZNd#ZxCkTb” data-sfc-root=”ep” jsuid=”jGM9nf_4″>1. Combine Your Passes

Fusion passes standard RGBA channels by default. To add extra layers (like Depth or Vectors) into the same stream:

• Add a Channel Boolean node: Branch your main Beauty pass into a Channel Boolean node (<code jsaction=”” jscontroller=”jM7ape#mDCHfd” data-sfc-root=”ep” jsuid=”jGM9nf_a”>Ctrl + Spacebar > type <code jsaction=”” jscontroller=”jM7ape#mDCHfd” data-sfc-root=”ep” jsuid=”jGM9nf_b”>Channel Boolean).
• Connect the Extra Pass: Connect your secondary pass (e.g., Z-Depth) into the secondary input (green foreground pip) of the Channel Boolean.
• Map the Channels: In the Channel Boolean inspector, go to the Aux Channels tab. Change the target channel (e.g., Z Buffer) from <em jsaction=”” jscontroller=”ABoAEd#jBqpid” data-sfc-root=”ep” jsuid=”jGM9nf_i”>Operation: Copy and select the corresponding channel from your secondary input.

<b role=”heading” data-sfc-cp=”” jsaction=”” jscontroller=”gjwZNd#ZxCkTb” data-sfc-root=”ep” jsuid=”jGM9nf_j”>2. Add and Configure the Saver Node

• Connect the Saver: Connect the output of your final Channel Boolean node to a Saver node.
• Set Format to OpenEXR: In the Saver properties, click Browse, name your file with a sequence extension (e.g., <code jsaction=”” jscontroller=”jM7ape#mDCHfd” data-sfc-root=”ep” jsuid=”jGM9nf_r”>Multilayer_Render.0000.exr), and ensure the format is set to OpenEXR.
• Enable Extra Channels: In the Saver inspector panel, look for the Channels or Format settings. Ensure that checkboxes for auxiliary channels (like Z, Coverage, or custom channels) are enabled so they write into the final EXR payload.

<b role=”heading” data-sfc-cp=”” jsaction=”” jscontroller=”gjwZNd#ZxCkTb” data-sfc-root=”ep” jsuid=”jGM9nf_z”>3. Choose the Right Compression

Multi-layer EXR files can get very large. In the Saver node’s format options, select an appropriate compression method:

• ZIP (Scanline or Block): Excellent for general 2D compositing and widely compatible across software like Nuke or After Effects.
• PIZ: Great lossless compression for noisy or grainy 3D renders, though it can be slightly slower to decode.

<b role=”heading” data-sfc-cp=”” jsaction=”” jscontroller=”gjwZNd#ZxCkTb” data-sfc-root=”ep” jsuid=”jGM9nf_16″>4. Render the Sequence

• Go to the top menu bar.
• Select Fusion > Render All Savers (or press <code jsaction=”” jscontroller=”jM7ape#mDCHfd” data-sfc-root=”ep” jsuid=”jGM9nf_1c”>F7).
• Set your frame range and click Start Render.

Gemini got the information from these two YouTube videos:

https://youtu.be/aiJB6YwtkWk?si=Ibc5VKvGf1pMsbQi

https://www.youtube.com/live/WWykNgpHAfc?si=zyByyjS30KaAFIxF

Hope this helps?

Please do come back and share your experiences, as Blender is making its way into the professional world of media production. And it is nice to know how it works in a Davinci workflow.

Atb
Mads

Some contents or functionalities here are not available due to your cookie preferences!

This happens because the functionality/content marked as “Google Youtube” uses cookies that you choosed to keep disabled. In order to view this content or use this functionality, please enable cookies: click here to open your cookie preferences.

[Mads Nybo jørgensen]

Hey Mark,

What a pain.

Are you able to roll back PPro to the previous version, or are locked-in on the new features?

What happens if you setup a new project, and import the Sequence (TimeLine) and try again?

Next I would try is to dissable the Morphcut, if you use it anywhere, and see if that will make the export happen?
If not, maybe dissable the Firefly clips?

I know this is counter productive, but if you a version exported, even with black holes in it, may be those individual clips will allow you to export those out one at a time, and re-insert on your exported (master) file.

Because of error messages, I did late last year go through a painfull experience with PPro on a feature film, where I had to delete all the tracks, except the one which I exported out.
Eventually all one at a time.
In the end, I had to export ONE voice track, starting 2 mins at a time, building up to 10 mins chunks, before getting it all out, and reassemble the exported files in a new project.

Adobe doesn’t really seem to care about stability, which is a shame.

Hope this helps?

Atb
mads

[Mads Nybo jørgensen]

Hey Johnny,

Are you slowing down and speeding up the audio too?

Or is it just the video?

Don’t know if you have mentioned it in another post, but it would be helpful to know what version of VEGAS you are working on, and the codec you have recorded your gameplay in?

Some lower-end codecs are Long-GOP, which I call “never-the-same-timecode-twice”.
In other words, depending on the length of the clip, it can change your in and out points.
I am not sure how VEGAS deals with this, but it is a big issue on Premiere Pro the longer your recordings are, and if you are using a fast cut pace.
Allegedly, DaVinci, FCP, and Avid have another way of dealing with Long-GOP than PPro, so they don’t have the same issue. Don’t know what the situation is with VEGAS.

Hope this helps?

Atb
Mads

[Mads Nybo jørgensen]

Hey Rob,

What kind of project is it?

Talking heads, music video…?

There are loads of tools out there that can find you a similar-looking video, but accuracy is difficult as they need to look at multiple frames to make an accurate match.

Although not entirely hopeful, You could try PPro’s Media Intelligence tool:
https://helpx.adobe.com/uk/premiere/desktop/organize-media/file-organization/search-for-media-using-ai-powered-media-intelligence.html

Whereas if it is talking heads, you could use Adobe to transcribe all of your rushes, and it will identify the clip based on the spoken word.

Should add that his has been a problem to solve since the early days of A.I. in Post Production.
One article here:
https://www.postmagazine.com/Publications/Post-Magazine/2019/November-December-2019/Outlook-2020-How-AI-can-automate-video-conform-c.aspx

Google Gemini keeps returning to Limecraft as a solution:
https://www.limecraft.com/

Maybe if you “chop up” your master and feed the individual clips in to a bin in Premiere Pro, the Adobe Media Intelligence might find both clips, and then you do a manual synch from there?

Good Luck

Atb
Mads

[Mads Nybo jørgensen]

Hey Chris,

eBay might be a good place to start.

Depending on where you are based, you might be able to get a local transfer facility to transfer from DV to Digital file for you.

Atb
Mads

[Mads Nybo jørgensen]

Hey Denise,

It is attractive to consider using A.I. for everything, but that won’t neccesarily solve the basic problems on a production.

Yes, you can use A.I./3D software for advanced low-cost story boards. But if you as a film-maker can not visualize the end result, how will you be able to go on set and give directions, or operate the kit?

This is where experience, and plenty of trial and error, comes in to the frame.

The first thing to go is the eyeline, particularly where you film 2 or more people in single shots, speaking to one another.

Will we see cameras that can tell you when something doesn’t match up, undoubtedly, or will it be cheaper to ask A.I. to fix it?

Meanwhile, Christopher Nolan’s Odessey was shot on IMAX 65mm film.

That is very expensive to make mistakes on, hence why a crew at that level will be amongst the best in the world.

As humans, we can only aspire to be better, although it won’t stop us using A.I. as a utility, but still with us in control.

Right?

Atb
Mads

[Mads Nybo jørgensen]

Hey Denise,

Good point about always making the test first.

However, over one hundred years of film-making, good cinematography (DOP/Lens/Lighting/Angle) can make most ordinary scenes look amzing. Or at least give enough space for the actors to do their bit.

Two good actors, dressed accordingly to what the script wants them to be. With a couple of props like good fancy coffee cup, fountain pen and expensive paper on what looks like a solid upmarket table. Then the viewer will soon have the impression of an expensive meeting room.
With a bit of editing, like an outside shot of an “imposing” building as a setup for the scene, then you can create an fancy home office without ever having one of those.

Although, back in the day when I was starting out, we did phone the town’s mayor and ask to use his office. Even got him to do a mock wedding for us (Yes, I got married without getting married, as a spotty teenager).

All you have to do, is ask.

And that is one curious courtesy that no amount of prompting an A.i. can “teach”.

What always fail, is lack of vision and planning.
After that, it is not knowing your eyelines, and how to shoot for them.

Having said that, there is nothing wrong with trying out some of the plenty different digital tools to get what you want.

Would be nice to get an update from @dovletov on how he is getting on with his project?
Or hopefully get to see his masterpiece.

Atb
Mads