Decent 240p to 4K upscaler with 1:1 speed?

[videomaniac]

Which software will be the most suitable for my case?
Ryzen 3600
32gb ram with 1TB SSD NVME
GTX960

upscaling:
240p Zoom video to 4K
720p talking head video to 4K

1:1 speed if possible (i.e. 5 mins to upscale a 5min video, otherwise worst case scenario = 20mins)

I'm not expecting magic, so just recommend the best you can, thanks!

[tebasuna51]

...so just recommend the best you can...

Please read the forum rules:

12) How NOT to post on this forum:
...
Do not ask "what's best" because this question cannot be answered objectively. Each and everyone has their own view about what's best in a certain area. The best is what works best for you!

The short answer to your post title:
"Decent 240p to 4K upscaler with 1:1 speed?"
A good TV.

[Atak_Snajpera]

The best upscaler has CSI. I highly recommend it!

[ognirats]

Which software will be the most suitable for my case?
Ryzen 3600
32gb ram with 1TB SSD NVME
GTX960

upscaling:
240p Zoom video to 4K
720p talking head video to 4K

1:1 speed if possible (i.e. 5 mins to upscale a 5min video, otherwise worst case scenario = 20mins)

I'm not expecting magic, so just recommend the best you can, thanks!

Not 1:1 but you can use nnedi3_rpow2. It does a pretty good job

[StainlessS]

The best upscaler has CSI. I highly recommend it!
Yep, amazing what they can do on CSI, nearly as good as in Blade Runner.

Google’s latest upscaling technology makes CSI TV shows a reality:- https://www.diyphotography.net/googles-latest-upscaling-technology-makes-csi-tv-shows-a-reality/

Google AI Blog :: High Fidelity Image Generation Using Diffusion Models [Friday, July 16, 2021 ]:- https://ai.googleblog.com/2021/07/high-fidelity-image-generation-using.html?m=1

Alternatively, diffusion models, originally proposed in 2015, have seen a recent revival in interest due to their training stability and their promising sample quality results on image and audio generation. Thus, they offer potentially favorable trade-offs compared to other types of deep generative models. Diffusion models work by corrupting the training data by progressively adding Gaussian noise, slowly wiping out details in the data until it becomes pure noise, and then training a neural network to reverse this corruption process. Running this reversed corruption process synthesizes data from pure noise by gradually denoising it until a clean sample is produced. This synthesis procedure can be interpreted as an optimization algorithm that follows the gradient of the data density to produce likely samples
Some of the example images @ above link look way too good to be true.

EDIT:
I'm not expecting magic
Well above AI Blog link looks a lot like magic. [but when available ???]

EDIT:
promising sample quality results on image and audio generation.
Arh, say "Image", not video, so maybe not too promising for video.

[Selur]

Some of the artifical learning methods are good at upscaling (nowhere near 'High Fidelity Image Generation Using Diffusion Models'), but they are far away from real time.
Also the result depends a lot on the input. -> no example input links?

[VoodooFX]

Decent 240p to 4K upscaler with 1:1 speed?

Try them yourself: http://avisynth.nl/index.php/Resize