Trimmed render file is bigger than original file

I did some cutting of a video, about 40% of its length, to remove some unwanted scenes.

A.
The rendered file size is about 800 MB while the original size is 300 MB. I assume that is due to the higher screen resolution I used in the rendering. Right?

B.
Original resolution is 1200 by 720, while rendered is 1920 by 1080. How does it give a higher screen resolution than original, by extrapolation?

Thanks.

What codec? We need a lot more information here.

Thanks.

Both original and rendered files are MP4.

B.
Original resolution is 1200 by 720, while rendered is 1920 by 1080. How does it give a higher screen resolution than original, by extrapolation?

Thanks.

Think it should be "interpolation" rather than "extrapolation". Thanks.

Did you activate SuperScale?
Of course, it'll be bigger by a factor of 2.4, but by how much depends on the chosen bitrate too.

I assume that is due to the higher screen resolution I used in the rendering. Right?

Actually, no. Resolution has no impact on file size. The formula is File Size = Bitrate x Duration.

The larger file, despite being shorter in Duration, has a higher Bitrate.

How does it give a higher screen resolution than original

Because you set it up that way. If you don't know how you did that...

https://www.blackmagicdesign.com/produc ... e/training

I assume that is due to the higher screen resolution I used in the rendering. Right?

Actually, no. Resolution has no impact on file size. The formula is File Size = Bitrate x Duration.

The larger file, despite being shorter in Duration, has a higher Bitrate.

Noted and thanks.

New to video editing and pardon me for my "stupid" question. The bitrate is influenced by the amount of data it needs to send over per scene? Thus resolution (data to send across) affects the file size?

Cheerio.

How does it give a higher screen resolution than original

Because you set it up that way. If you don't know how you did that...

https://www.blackmagicdesign.com/produc ... e/training

Thanks.

Sorry for my poor command of English, I did not phrase my question correctly.

Yes, know that I can set different resolutions, 720, 1080, 4k etc.

My intended question is, to change from 1200 by 720 to 1920 by 1080, how/where does it get the extra resolution from width 1200 to 1920, and height 720 to 1080?

AI algorithm to analyse adjacent points, come up with the best estimate of the color in between these 2 points and then insert these points in the scene?

Cheerio.

There is always interpolation, i.e. 'inventing' new pixels, but the algorithms are different.
They range from simple ones, working in different ways to average information from neighbouring pixels, to complex (and far slower) ones like SuperSample, which claims some sort of neural processing.
But the number of pixels to compress will always increase when upscaling.

There is always interpolation, i.e. 'inventing' new pixels, but the algorithms are different.
They range from simple ones, working in different ways to average information from neighbouring pixels, to complex (and far slower) ones like SuperSample, which claims some sort of neural processing.
But the number of pixels to compress will always increase when upscaling.

Noted and thanks.

Any advice on how much upscaling is ok without making the scene unrealistic and thus a waste of time doing it? 10% or even 30%? Just a rough guide.

Thanks.

SuperScale looks pretty good, even at a 30% increase, but all of this is highly scene-dependent. And GIGO always applies…

Noted and many thanks!

The bitrate is influenced by the amount of data it needs to send over per scene?

Typically, no. The bitrate is normally set by the user.

The exception to that rule is with Constant Quality encoding options, which aren't as common as I would like. With CQ, you set the target quality and the software will use as many bits as it needs to hit that target. It's the best way to encode things, I believe.

But normally you're setting a target bitrate, which will not be as influenced by things like frame rate and resolution like CQ mode is.

But with all encoding options, the formula is the same.

File Size = Bitrate x Duration. Files are made of bits. Bits per second multiplied by the number of seconds is the total number of bits - the file size.

how much upscaling is ok

Depends on how it's done. Normally I would advise "none". You should never upscale a clip.

But I agree with Uli. SuperScale is a good exception to that rule.

Another user claims that the following is even better than SuperScale.

https://topazlabs.com/video-enhance-ai/

It is. But extremely slow. Like sloooooow.