Copy link to clipboard
Copied
Hello,
I recently completed a project and my client is dissatisfied with results of the quality of the exported project compared to the color, clarity, and quality of the raw footage.
Here are screen captures for comparison of the raw source, the exported version, and the two side-by-side
Here is a shot of the raw footage:
Here is a shot of the same frame, exported:
And two examples, side-by-side:
Here are my export settings on Premiere Pro:
Format: H.264 (.mp4)
Framerate: Match the source video’s framerate
Frame size: 3840x2160
Field Order: Progressive
Aspect: Square Pixel
Performance: Hardware Encoding
Profile: High
Level: 5.2
Copy link to clipboard
Copied
First it's helpful to get the nomenclature correct. "RAW" is a specific thing ... it's a particular encoding process of the camera. And is not just 'the original state' of the clip.
So the original image format/codec would be somewhat helpful ... though that it's 8 bit tells us you can't do much to the image before it breaks up. If they want better quailty after post processing, they need to get out of 8 bit capture. Period.
There's very little to be done with 8 bit files, and I've a LONG history of making them ... less bad. I've done "near miraculous!" recovery for other users and clients.
The first thing is you're dealing with long-GOP media and exports. Those use blocks of similar-data image chunks to save space. By eliminating the differences between nearby similar color pixels in blocks. Making a group of 4, 6, or 9 pixels all the same. This saves the amount of data to be written about that image. It throws out detail by the scad, though.
With 10 bit media, you don't notice it that much ... always. With 8 bit media, it's gonna bite hard nearly instantly when you do anything either tonal or chroma to the image.
Do you see why you start seeing image issues fast now?
So to get as much as possible, you do only very small changes, and to get the total desired change, you probably need to break up the desired change across several tools. They all use different math, so this means the change is applied in a way that has a better chance of getting through compression without macro-blocking the image.
So ... to darken parts, you would do a titch of say Wheels luma slider down, and something from the Basic tab on that same tonal range, and maybe just touch the RGB curve.
I would use a different instance of Lumetri for each type of work. Stacked on the clip directly. So one for working the shadow tonality, another for the highlight tonalit, and after the tonal work, then one for shadow color controls, another for highlight color, and a last one for any overall color past that.
The key is that you only do a little bit with each tool. Never enough to 'fix' the image, and between them, make it a bit better. Don't try for perfection, just ... visibly better.
Then if it survives a test export, you've done it.
Sometimes you need to first export to say ProRes 422, re-imported to Premiere, then make your H.264/5 export from that clip. It can at times get less macro-blocking.