Guides How to use foto2vam

[CaveGentleman]

To see the output, just run it from a command prompt or PowerShell instead of from windows explorer. Having said that, the most likely culprit, if you look in your input folder, you should see output files for each image you tried to use. I'll bet that it didn't like one or more of them

Thanks enigma. I'll give it a try.

 

[CaveGentleman]

To see the output, just run it from a command prompt or PowerShell instead of from windows explorer. Having said that, the most likely culprit, if you look in your input folder, you should see output files for each image you tried to use. I'll bet that it didn't like one or more of them

So I couldn't figure out how to run it from CP or PS, but I did redownload it and copied an input folder from a previous (successful) project, and that worked. What's strange is that is that I have been unsuccessful with every single new attempt.---and I have tried many new models with various photo qualities.

TL;DR
I think your right that has something to do with the images. Otherwise the old images wouldn't have worked.


*UPDATE*
It started working again with new projects after I saved the input images as jpegs instead of pngs. This doesn't make sense, since most of my previous projects were done usings pngs; but I'll accept the workaround for now.

 

[dwda]

is there going to be an updates of foto 2 vam?

 

[ehowin]

great tutorial, thank you. Very nice app.

 

[redviking]

Is the a way to change the base model, I tried to do it myself but it gave me an error.

 

[hazmhox]

Is the a way to change the base model, I tried to do it myself but it gave me an error.

You could. You'll have to redo the whole training process and modify all potential morphs used.
Never done that since the training process is super long and not documented enough.

 

[ftvam]

Helllo!
I can't figure out how to use created *.json file in the application? in which section?
i put it into VaM_Updater\Custom\Atom\Person\Morphs\female and what next?

thank you!

 

[hazmhox]

Helllo!
I can't figure out how to use created *.json file in the application? in which section?
i put it into VaM_Updater\Custom\Atom\Person\Morphs\female and what next?

thank you!

Hi!
Let me quote the documentation :

Open it in VAM, add some light... Load your scene.

This is not a morph, the result of foto2vam is just a simple scene you load.

 

[ftvam]

Hi!
Let me quote the documentation :

This is not a morph, the result of foto2vam is just a simple scene you load.

Got it now. Thanks a lot!!

 

[bboyneko]

Hi, I see in the release there is a test models folder. I tried to place those files into the /model directory, expecting the program to output a bunch of different vams based on the models. But instead, it only generates one file.

What am I doing wrong? Is there any benefit of trying different outputs based on different models from that test models folder? Thanks!

 

[hazmhox]

Hi, I see in the release there is a test models folder. I tried to place those files into the /model directory, expecting the program to output a bunch of different vams based on the models. But instead, it only generates one file.

What am I doing wrong? Is there any benefit of trying different outputs based on different models from that test models folder? Thanks!

The model is the one used to generate the final output. I'm pretty sure that changing it might be a problem. That's all I can say (I haven't created the software, it is only a redistribution of the original code).

 

[barnaby69]

Hi, thanks for the tutorial and links. Sadly, I get an error message every time. I have tried the version from the download link at the top of the tutorial, and I have also tried downloading both 1.10 & 1.11 from Mega. Same error occurs in all 3 versions. Any advice would be greatly appreciated! Thanks.

C:\F2V1.11>foto2vam.exe
Traceback (most recent call last):
File "Utils\Face\encoded.py", line 4, in <module>
ModuleNotFoundError: No module named 'face_recognition_hst'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "foto2vam.py", line 5, in <module>
File "c:\users\chris\envs\foto2vam_staging\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 627, in exec_module
File "Tools\CreateTrainingEncodings.py", line 3, in <module>
File "c:\users\chris\envs\foto2vam_staging\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 627, in exec_module
File "Utils\Face\encoded.py", line 6, in <module>
File "c:\users\chris\envs\foto2vam_staging\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 627, in exec_module
File "site-packages\face_recognition\__init__.py", line 7, in <module>
File "c:\users\chris\envs\foto2vam_staging\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 627, in exec_module
File "site-packages\face_recognition\api.py", line 4, in <module>
ImportError: DLL load failed: A dynamic link library (DLL) initialization routine failed.
[9708] Failed to execute script foto2vam

 

[enigma]

Hi, thanks for the tutorial and links. Sadly, I get an error message every time. I have tried the version from the download link at the top of the tutorial, and I have also tried downloading both 1.10 & 1.11 from Mega. Same error occurs in all 3 versions. Any advice would be greatly appreciated! Thanks.

C:\F2V1.11>foto2vam.exe
Traceback (most recent call last):
File "Utils\Face\encoded.py", line 4, in <module>
ModuleNotFoundError: No module named 'face_recognition_hst'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "foto2vam.py", line 5, in <module>
File "c:\users\chris\envs\foto2vam_staging\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 627, in exec_module
File "Tools\CreateTrainingEncodings.py", line 3, in <module>
File "c:\users\chris\envs\foto2vam_staging\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 627, in exec_module
File "Utils\Face\encoded.py", line 6, in <module>
File "c:\users\chris\envs\foto2vam_staging\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 627, in exec_module
File "site-packages\face_recognition\__init__.py", line 7, in <module>
File "c:\users\chris\envs\foto2vam_staging\lib\site-packages\PyInstaller\loader\pyimod03_importers.py", line 627, in exec_module
File "site-packages\face_recognition\api.py", line 4, in <module>
ImportError: DLL load failed: A dynamic link library (DLL) initialization routine failed.
[9708] Failed to execute script foto2vam

what version of python do you have installed? mine is 3.6.5 (64bit). that's all I can think of that would prevent you from running it against the included samples (which I recommend before you start adding your own images)

 

[barnaby69]

what version of python do you have installed? mine is 3.6.5 (64bit). that's all I can think of that would prevent you from running it against the included samples (which I recommend before you start adding your own images)

Thanks for the suggestion, but I do have Python 3.6.5 64bit installed, same as yourself. Also I have tried running it just using the supplied samples without adding any additional pictures, still causes this error message. It appears to be an absolute path hardcoded into the Python somewhere, as shown in the error message, obviously I don't have a "c:\users\chris\" folder on my PC. I don't really follow how the build can work for anyone else other than the original person who compiled it, but I guess I am missing the point. I am not particularly technical but the guide is fairly easy to follow and states this should just run as is from the .exe once downloaded. I followed a different guide to try to build/compile my own version from the github source code, lets just say that didn't go well at all ?

 

[barnaby69]

Thanks for the suggestion, but I do have Python 3.6.5 64bit installed, same as yourself. Also I have tried running it just using the supplied samples without adding any additional pictures, still causes this error message. It appears to be an absolute path hardcoded into the Python somewhere, as shown in the error message, obviously I don't have a "c:\users\chris\" folder on my PC. I don't really follow how the build can work for anyone else other than the original person who compiled it, but I guess I am missing the point. I am not particularly technical but the guide is fairly easy to follow and states this should just run as is from the .exe once downloaded. I followed a different guide to try to build/compile my own version from the github source code, lets just say that didn't go well at all ?

So I have figured out the problem after going through some reddit threads. Turns out the processor in the desktop PC I was using is too old and missing some instruction sets. I was using an old desktop with an Intel Q9650 chip. The message didn't really help point to this as the issue, but seeing that others had experienced issues I tried to run it on my newer gaming rig with a newer AMD chip and hey presto, no problem! Just though I would post in case anyone else comes here looking for the solution - try using a processor which isn't a museum piece ?

 

[mehbebe]

It looks like the latest version of this (4 from what I can see) does not work with the latest version of tensor flow (2.0).
I was able to get it working however by modifying 2 .py scripts.
Tools - Train.py & MakePrediction.py
The following lines need to be modified in the code:
tf.ConfigProto
change to
tf.compat.v1.ConfigProto

tf.Session
Change to
tf.compat.v1.Session

Was quite frustrating having to learn python script just to fix a couple lines of depreciated code.

 

[hazmhox]

Was quite frustrating having to learn python script just to fix a couple lines of depreciated code.

I had to fix some stuffs prior to this release too. (the old python one)
To be honest, the stuff is gonna die sooner or later, since it's not supported by the original creator.

And now I'm at a point where I just see this as "worst" than a manual iteration on the morphs due to the results you can get. It never had been really trained a lot... if the overall training samples was way bigger and accurate, this could have made an incredible way to create a base model.
The reality is far from that : )

Thank you for the report for people who'd like to try the python version!

 

[mehbebe]

I had to fix some stuffs prior to this release too. (the old python one)
To be honest, the stuff is gonna die sooner or later, since it's not supported by the original creator.

And now I'm at a point where I just see this as "worst" than a manual iteration on the morphs due to the results you can get. It never had been really trained a lot... if the overall training samples was way bigger and accurate, this could have made an incredible way to create a base model.
The reality is far from that : )

Thank you for the report for people who'd like to try the python version!

Yeah I was thinking the same thing. When it comes to "training" a neural network on a face, I'm used to the concept requiring a whole lot of images from all sorts of angles, and different expressions etc, so it can really figure out what the face looks like, and i'm used to "training" taking hours of time, but foto2vam seems to be done within a couple of seconds, so its almost as if there is no training involved.
The results for me were quite mixed, but most of the time it seems to capture certain features and over do those features to a point it can look cartoonish.
Sometimes the results can be alright though.

 

[hazmhox]

Yeah I was thinking the ...

Actually, the training is the long part. But if you base your generation on your trained model, the software can produce results in a couple of seconds. The actual flaw in foto2vam is the model, it is not trained (or detailed) enough to produce good results. It was great and fun at the beginning, but... it gets old very fast.

And yes, I agree, really cartoonish... but the worst : extremely similar results even with extremely different sources.

 

[lj92458]

generated.photos
this website is too slow, and Prone to collapse.

 

[Picarelli]

I tried with the below image example from imgur , but my results using just 'Output'(not merged) seem off and others I've tried have the same off-ish quality. Not sure if you're able to replicate the quality of the better example or what the trick(s) are. Not sure if its related to the base model structure/morph limits - the nose always seems bulbous and jawline more masculine, etc.

Thanks

View attachment 11637View attachment 11638

View attachment 11635View attachment 11636w
[/CITAÇÃO]
ola poderia me dar uma ajuda como instalar o programa foto2vam eu clico no exe do programa não acontece nada

 

[Picarelli]

ola poderia me dar uma ajuda como instalar o programa foto2vam eu clico no exe do programa não acontece nada que alguém pode me ajudar por favor

 

[hazmhox]

ola poderia me dar uma ajuda como instalar o programa foto2vam eu clico no exe do programa não acontece nada que alguém pode me ajudar por favor

You'll have to make the effort to use google translate and use english if you want an answer! Thank you!

 

[Picarelli]

You'll have to make the effort to use google translate and use english if you want an answer! Thank you!

ola could give me a help how to install the program foto2vam I click on the exe of the program does not happen anything that someone can help me please how to install the program?

 

[hazmhox]

ola could give me a help how to install the program foto2vam I click on the exe of the program does not happen anything that someone can help me please how to install the program?

Ola!

The exe should work pretty much directly after putting it in the folder. It is hard to help if you without errors, screenshots, and more details.