Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
|
|
Thread Tools | Search this Thread | Display Modes |
15th January 2018, 18:00 | #1 | Link |
I'm Siri
Join Date: Oct 2012
Location: void
Posts: 2,633
|
or maybe we dont need GANs to generate photo realistic image upscales..
I went on and now experimented on 1/9 of my super resolution neural net, 5 million parameters roughly trained out of total 45 million parameters, I only trained it with 3 actual images (augmented to 18 images via flipping, mirroring and stuff) for 20 epochs of backprop, then tested the neural net on 4 images that it had never seen before
test results small: upscaled: small: upscaled: small: upscaled: small: upscaled: the results already look pretty photo realistic to me, I aint got no GAN in my neural net, no perceptual loss either, I simply picked L2 (aka MSE) as the loss function (PSNR friendly) and I honestly wasn't expecting something this good, the results are almost as photo realistic as the results of GANs, and the chance of shit going south should be much smaller than GANs since the neural net was optimized for L2, which is very loyal to the ground truth,,, so maybe we don't need GANs to have photo realistic upscales, all we need is a better neural net structure (the structural design of my neural net serves a special physical significance, it's not about brainlessly stacking layers and blocks like some other models) |
15th January 2018, 18:32 | #2 | Link |
Registered User
Join Date: Sep 2007
Posts: 5,374
|
the textures don't quite look right , keep on training it f2 . 5M not enuf
It's progress and it looks good, but there is a "coarseness" to the fine details that is reminiscent of nnedi2/3 upscaling artifacts . The fine details are what differentiates this approach There are some issues with the bridge /railing that weren't there on your earlier example (look almost like field/interlace artifacts) Last edited by poisondeathray; 15th January 2018 at 18:49. |
15th January 2018, 18:55 | #3 | Link | |
I'm Siri
Join Date: Oct 2012
Location: void
Posts: 2,633
|
Quote:
well, it doesn't look quite rite for mainly 2 reasons, and the number of parameters is actually not one of them, first, training set of 3 effective images is too small, also, 20 epochs of backprop aint the shit for a large neural net to converge, I'd say probably at least 300 epochs |
|
15th January 2018, 19:00 | #4 | Link |
Registered User
Join Date: Sep 2007
Posts: 5,374
|
I don't know the details of what it takes to train... Can you pause and restart for example ? Does it take 1 contiguous session ? Can you train it overnight at school for example for a few weeks ?
It looks good, I just think you can do better Any thoughts as to why would this version generate those bridge horizontal artifacts ? |
15th January 2018, 19:14 | #5 | Link | |||
I'm Siri
Join Date: Oct 2012
Location: void
Posts: 2,633
|
Quote:
Quote:
Quote:
|
|||
15th January 2018, 19:33 | #6 | Link | |
I'm Siri
Join Date: Oct 2012
Location: void
Posts: 2,633
|
Quote:
no, this is not what I would call, photo realistic |
|
15th January 2018, 19:52 | #7 | Link | |
Member
Join Date: Jan 2018
Posts: 17
|
Always enjoy seeing NN image scaler examples. Looking great.
Quote:
|
|
15th January 2018, 20:21 | #8 | Link | |
Registered User
Join Date: Sep 2007
Posts: 5,374
|
Quote:
This version has some coarseness features too (definitely better than nnedi3), but going back now I can see why - the other pics in the other thread were at lower dimensions So I'm thinking you can improve on it |
|
16th January 2018, 17:20 | #11 | Link |
I'm Siri
Join Date: Oct 2012
Location: void
Posts: 2,633
|
there's zero point to compare this with NGU just yet, cuz it's doomed to introduce hell lot more artifacts than NGU for now, cuz the neural net is not yet fully trained, actually make that, yet just "barely trained"
anyways, the castle image, the source image is way too soft and doesn't have a lot of details,, so the result is just mediocre quality and suffers from mild level of artifacts the cartoon image is 100% pointless, it's not photo realistic to begin with the result of lighthouse image suffers from crappy artifacts.. Last edited by feisty2; 16th January 2018 at 17:22. |
16th January 2018, 18:32 | #13 | Link | |
I'm Siri
Join Date: Oct 2012
Location: void
Posts: 2,633
|
Quote:
it works for videos (temporally stable), but it's slow |
|
16th January 2018, 18:34 | #14 | Link |
Formerly davidh*****
Join Date: Jan 2004
Posts: 2,496
|
feisty2, can you make anything of this?
https://github.com/DmitryUlyanov/deep-image-prior Seems to be about using untrained neural nets for noise reduction, inpainting, and upscaling. One of the test images from this thread is in the paper. |
16th January 2018, 18:41 | #15 | Link | |
Registered User
Join Date: Sep 2007
Posts: 5,374
|
Quote:
Looks promising too Does it lend itself more easily to vapoursynth since both use python ? |
|
16th January 2018, 19:06 | #16 | Link | |
I'm Siri
Join Date: Oct 2012
Location: void
Posts: 2,633
|
Quote:
there're also other multi purpose neural nets based on supervised learning https://arxiv.org/pdf/1608.03981.pdf https://arxiv.org/pdf/1606.08921.pdf |
|
16th January 2018, 19:21 | #17 | Link | |
I'm Siri
Join Date: Oct 2012
Location: void
Posts: 2,633
|
Quote:
and amd gpus r fucked too, all require nvidia gpus python does rule the machine learning field, so basically implementations for all recent papers r coded in python |
|
16th January 2018, 20:37 | #18 | Link |
HeartlessS Usurer
Join Date: Dec 2009
Location: Over the rainbow
Posts: 10,980
|
Respect master Feisty, for something so early in development, is looking really good.
__________________
I sometimes post sober. StainlessS@MediaFire ::: AND/OR ::: StainlessS@SendSpace "Some infinities are bigger than other infinities", but how many of them are infinitely bigger ??? |
16th January 2018, 21:10 | #19 | Link | |
Registered User
Join Date: Sep 2007
Posts: 5,374
|
Quote:
No 1 clicky, or even semi-easy solution for a script Back to training then... |
|
Thread Tools | Search this Thread |
Display Modes | |
|
|