The ebsynth's discuss from jamriska

Question : Hi, is it the open source version of ebsynth app ? (https://ebsynth.com/)

Hi, is it the open source version of ebsynth app ? (https://ebsynth.com/)

If it is, what is the equivalent for video and keyframes in ebsynth ? Because when I tried it, the result is not quite the same

Thanks

How to get the finished pngs

Hi!

I've just started with EbSynth, but I can't get the finishe pngs. I followed the tutorial that comes with the download, I've followedkrisfire's tutorial (one of the firsts that allways shows up for me), I've followed Joel HAver's that says I have to sort my original video pngs into folders according to the keyframes I have, I've also done it all with the example that came with the download, and still, whenever I hit the "Synth" button, I'll get the new folder, but with nothing in it. I really don't know what to do, I'm still hoping it's a silly thing I forgot, but it's been almost 10 hours now and I'm quite desperate....

consider adding Mask option?

Hi Ondrej,

Great project. I see the executable version of Ebsynth has this mask option, with which you can use external generated mask for inpainting style transfer. Do you consider adding that feature to the repo?

Thanks,
Tony

plan to update the code with the public GUI version?

thanks for the great tool.
any plan to upgrade the code with the GUI version?

simple example OK, but my photo fail

OK, this is my first (edgy) try beyond the examples and I know doing support is not PhD cup of tea :)
I just want to understand what is wrong with my settings. Thank you.

I took a 360 picture that includes me and someone else taking the picture.

I'd like to clone some parts onto others so I made guides:
a source:

and a target:

Yellow gradient is the same (like gradient on the rock example), then the other colors are for cloning guides.

Of course, everything matches and pics are same res.

I added lines onto the green mask to help follow the lines, like positioning guides within segmentation guides.

Result is wrong everywhere, even yellow gradient:

Pictures are the same size.
I also tried without curved guides and without the masks in the bakground (cars).
I even tried with same source and target:

I tried with JPG and PNG, same kind of result, almost the same result in fact, for instance: no tree in all cases, while it was not masked.

What is wrong?
Or is it a specific PNG settings, resolution?
Thanks

fyi

https://twitter.com/karenxcheng/status/1565003860854185984?s=20&t=4_P3UToO3QdyMj1BxuRRQQ

Pre-rendered motion vectors?

Hi, not an issue but a question - if using this on CG content generated from a renderer, where the renderer can generate pixel-perfect 2d motion vectors, is it possible to use these pre-rendered motion vectors as inputs to EbSynth?

cheers

Wiki for community support

I notice you get questions about the operation of the program in the issues now. Is it an idea to open a wiki, to allow for community Q&A? I have been using your tool today for an animation and I think I can help answer some of the questions from the community.

Also you could add tips for specific use cases.
This is amazing work and I think you should sell this to adobe or something.
I cannot (yet) really figure out what you did, thought it was Neural Net, I was wrong, thought you used forward prediction algorithms from mpeg to distort textures incrementally, but it turns out I was wrong AGAIN.
Well done, anyway...

Compiling ebsynth_win64 for cuda?

I've tried changing the compute_30 field, which lets it compile, but the gpu output returns a black image, whereas the cpu output returns the output as expected.

Anyone have any ideas?

15 was unexpected at this time

Hi,
im running build-win64-cpu+cuda.bat and its immediately failing with "15 was unexpected at this time"
I guess its probably coming from this line: https://github.com/jamriska/ebsynth/blob/master/vcvarsall.bat#L6
if %vsver% geq 15 (

Im running windows 10 64 bits.
Thanks

How can I get source_Gseg.png

Hi, how can I get source_Gseg.png?

Code for generate G_seg?

Is there any implemented G_seg generator code?

Please reupload pre-compiled version

I tried with visual studio 2022 2019 and 2017

@jamriska

all fails

G:\Rerender_A_Video auto install\Rerender_A_Video\venv\Scripts>activate

(venv) G:\Rerender_A_Video auto install\Rerender_A_Video\venv\Scripts>cd..

(venv) G:\Rerender_A_Video auto install\Rerender_A_Video\venv>cd..

(venv) G:\Rerender_A_Video auto install\Rerender_A_Video>python install2.py
Build Ebsynth Windows 64 bit. If you want to build for 32 bit, please modify install.py.
.\build-win64-cpu+cuda.bat
"C:\Program Files\Microsoft Visual Studio\2022\Community\VC\Auxiliary\Build\vcvarsall.bat"
ebsynth.cpp
ebsynth_cpu.cpp
ebsynth_cuda.cu
C:\Program Files (x86)\Microsoft Visual Studio\2017\Enterprise\VC\Tools\MSVC\14.16.27023\include\vcruntime.h(184): error: invalid redeclaration of type name "size_t"

C:\Program Files (x86)\Microsoft Visual Studio\2017\Enterprise\VC\Tools\MSVC\14.16.27023\include\vcruntime_new.h(66): error: first parameter of allocation function must be of type "size_t"

C:\Program Files (x86)\Microsoft Visual Studio\2017\Enterprise\VC\Tools\MSVC\14.16.27023\include\vcruntime_new.h(71): error: first parameter of allocation function must be of type "size_t"

C:\Program Files (x86)\Microsoft Visual Studio\2017\Enterprise\VC\Tools\MSVC\14.16.27023\include\vcruntime_new.h(77): error: first parameter of allocation function must be of type "size_t"

C:\Program Files (x86)\Microsoft Visual Studio\2017\Enterprise\VC\Tools\MSVC\14.16.27023\include\vcruntime_new.h(82): error: first parameter of allocation function must be of type "size_t"

C:\Program Files (x86)\Microsoft Visual Studio\2017\Enterprise\VC\Tools\MSVC\14.16.27023\include\vcruntime_new.h(184): error: first parameter of allocation function must be of type "size_t"

C:\Program Files (x86)\Microsoft Visual Studio\2017\Enterprise\VC\Tools\MSVC\14.16.27023\include\vcruntime_new.h(199): error: first parameter of allocation function must be of type "size_t"

C:\Program Files (x86)\Microsoft Visual Studio\2017\Enterprise\VC\Tools\MSVC\14.16.27023\include\vcruntime_new_debug.h(23): error: first parameter of allocation function must be of type "size_t"

C:\Program Files (x86)\Microsoft Visual Studio\2017\Enterprise\VC\Tools\MSVC\14.16.27023\include\vcruntime_new_debug.h(31): error: first parameter of allocation function must be of type "size_t"

C:\Program Files (x86)\Microsoft Visual Studio\2017\Enterprise\VC\Tools\MSVC\14.16.27023\include\type_traits(168): error: class template "std::_Is_function" has already been defined

C:\Program Files (x86)\Microsoft Visual Studio\2017\Enterprise\VC\Tools\MSVC\14.16.27023\include\type_traits(212): error: class template "std::_Is_memfunptr" has already been defined

C:\Program Files (x86)\Microsoft Visual Studio\2017\Enterprise\VC\Tools\MSVC\14.16.27023\include\type_traits(1849): error: class template "std::result_of" has already been defined

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\crt/common_functions.h(117): error: first parameter of allocation function must be of type "size_t"

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\crt/common_functions.h(118): error: first parameter of allocation function must be of type "size_t"

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\crt/common_functions.h(240): error: first parameter of allocation function must be of type "size_t"

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\crt/common_functions.h(241): error: first parameter of allocation function must be of type "size_t"

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_20_intrinsics.hpp(108): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(104): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(105): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(109): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(110): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(111): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(112): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(113): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(114): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(115): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(116): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(117): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(118): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(119): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(120): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(122): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(123): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(124): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(125): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(126): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(127): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(128): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(129): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(130): error: asm operand type size(8) does not match type/size implied by constraint 'r'

c:\program files\nvidia gpu computing toolkit\cuda\v10.0\include\sm_32_intrinsics.hpp(131): error: asm operand type size(8) does not match type/size implied by constraint 'r'

Error limit reached.
100 errors detected in the compilation of "C:/Users/King/AppData/Local/Temp/tmpxft_00004498_00000000-18_ebsynth_cuda.cpp1.ii".
Compilation terminated.
FAILED
Failed to install Ebsynth.

(venv) G:\Rerender_A_Video auto install\Rerender_A_Video>

Colab notebook??

yk, for all the potato PC owners out there.... Cuz my MX250 isn't doing it......................
We can actually collaborate, I'm kind of a coder... At least I'll do my best.....

Verbose output during run

Seemingly for complex guides or large images the application can take some minutes to complete. It would be helpful to have a verbose option to see that it is still working, and has not crashed; and also to estimate how long it might take to complete, so that I can decide how elaborate a snack to prepare whilst I am waiting!

StyLit Knee Point

Hi,

Figure 6 of the StyLit paper shows a knee point finding step on the sorted errors of a reverse NNF. It looks like this implementation only ever uses regular NNFs (i.e. target-sized 2D arrays of indices in the source). Am I misunderstanding the implementation here? It seems like this implementation uses algorithm 1 from the StyLit paper, but never does the step in the paragraph that starts with "To complete the algorithm." I would appreciate any insights, thanks!

Image sequence

What parameters do I need to use in the CLI to use this with an image sequence? The precompiled tool on ebsynth.com is great but I'd like to use the CLI version with cuda

generator of Gpos

I don't have generate the distortion of squares deformation (Gpos)
It exists some software that do it?

There is no scroll bar in the window, so only about 10 keyframes can be added

I found a very serious problem, I added a lot of key frames, about 30, but the window did not provide scrollbars, resulting in only 10 key frames added.

SVG import and export

Svg files would produce better result for synthesizing lineart and colour images. It should be added as an import / export format

Max number of keyframes allowed

Hello,
We are currently testing your tool on a production environment and were wondering about how many keyframes are set to max (or if there's a way of recompiling allowing more of them) as currently we are unable to use more than 2 concurrent keyframes at the same time.

Thanks.

Problems when compiling for ubuntu 18.04

I run the build-linux-cpu_only.sh script, and the terminal just remains blank. I've tried using the windows version, but my windows PC is so slow. Please help fix this. I really need to start testing with ebsynth asap.

Can't compile cuda version for linux

Getting a lot of src/ebsynth_cuda_texarray2.h:76:55: error: invalid conversion from ‘const int*’ to ‘int’ [-fpermissive] errors when trying to compile with the build-linux-cpu+cuda.sh script.
I'm missing any requeriment?

Compiling with modern Linux / CUDA: -arch option

On running the Linux build script for cpu+cuda I got an error to the effect that "compute_30" wasn't recognised. Seemingly this has to do with specifying a minimum architecture which is now obsolete and should be met by any modern GPU / CUDA version. I just deleted the option in the compile script and it built and ran fine. Note that the compile options are also set to terse, so there will be no output during the compile, which takes a couple of minutes; please be patient (or set it to verbose).

Code to generate diffusion curves

Could you provide the code to generate diffusion curves and render it onto an image.

As far as I've looked, the authors of the original paper haven't released their source code.

Apple silicon/ARM/MPS

Can someone compile this for arm apple silicon

cudadevrt.lib error while compiling

Hello, I have both tried compute_89 and sm_89 for my enviroment but got this error everytime: "cudadevrt.lib" not found. Anyone had this or fixed this issue? I have RTX 4090 and CUDA 11.8 installed correctly. Visual Studio 2022.

AUR package doesn't comply

When i try to install ebsynth from aur package, it doesn't install. I am a linux and espacially arch linux newbie and i dont really know what to do. Can someone explain how to install it to linux step by step?

Generator gesp gpos

It exists something to generate it of way semi-automatic?

Where can I find some helpful documentation for linux usage?

Hello @jamriska , first of all, thanks and congratulations for this incredible work!
I was truly amazed when I saw a video showing the EbSynth software running this algorithm and I'm attempting to get used to the Linux interface of the program.
I've successfully built it in my LinuxMint, based on Ubuntu 18.04.2, but now I'm having trouble to understand what each parameter does and how to use them. I couldn't find any information about them neither on the README (and the other project files) or on the ebsynth website.
Am I missing something?
Thank you!

Pre-built Windows binary Failed

the Pre-built Windows binary : http://jamriska.cz/ebsynth/ebsynth-win64.zip Failed？ anyone can provide？thanks a lot!!!!

about "Special treatment of open mouth and eyes"

I read your paper about your special treatment for eyes, I am confusing about the real operation of sentence “To improve that, a mask with increased weight of Gapp (d) is used to guide the transfer of lighter texture areas of the style exemplar to the teeth” .In my opinion , Is that mean ：
1）add another independent guide channel, which is the mask of eye and mouth.
2) Use the mask(of eye and mouth) to matte the Gapp to produce a new guide channel with only eye and mouth in it, and the relevant weight is set higher than Gapp guide.
3) directly use the 0/255 mask of eye and mouth as a new guide.
which one is correct? or all wrong? I‘ll appreciate any replys, Thanks!!!!

Windows. ebs file edition for automatic filling ?

Hello!
Thanks for you awesome window implementation !
I was wondering is there was any way I could edit a .ebs file project to automatically fill first/last/keyframe/output via-python ? The manual process is quite tedious when dealing with long sequences and multiple keyframes.

Cheers !

code for ebsynth gui process with the cli

wondering if there a script publically available or if we can get a script shared that is similar to the eb synth gui but utilizes the cli and has the same output

de-flicker

Hi.
Is there a way to create "de-flicker" from the command line?

https://cf.mastohost.com/v1/AUTH_91eb37814936490c95da7b85993cc2ff/mographsocial/media_attachments/files/000/296/915/original/3dc88fa4b22b735b.mp4

How to use it for color transfer?

I am trying to use ebsynth for color transfer.

example:

image2 is style.
guide is only one: grayscale image2 -> grayscale image1
result is image3
it looks like image3 lost colors.

@jamriska any advice?

Question about Omega.

I appreciate you for sharing your work to public and your dedication.
Thanks to you, i could learn more about example-based image synthesize.

During the code study, i found calculation of the Omega in the patch matching.
Which was the concept that i could not find from patch match algorithm (PatchMatch: A Randomized Correspondence Algorithm for Structural Image Editing).
Can i know the concept of Omega. And if reference of that concept is exist, can i know the reference(paper)?

I look forward to your reply and thank you again.

error: the cuda backend is not available!

I installed ebsynth and the cuda toolkit. Right now it gives me the error above when I try to set the backend to CUDA. Anyone have any luck getting around this? I had to change the nvcc compiler options to compute_52 instead of compute_30 as others have pointed out but compilation worked and gave no errors.
Is there a version of CUDA that I should be using? Should I even bother or is the performance increase not worth it?

Ubuntu 21.04
GTX 970
Installed version of cuda is: cuda_11.2.r11.2

jamriska / ebsynth Goto Github PK

ebsynth's Issues

Recommend Projects

Recommend Topics

Recommend Org