MVE using Machine Video Editor to correct Face alignments and Landmarks - RE-EXTRACT

[kakks]

Why does it keep saying "no faces found"?  I have executed "5) data_dst faceset extract". and I got "aligned" and  "aligned_debug" folder. and then, I put alined,aligned_debug, frame folder into MVE just like tutorial says but it showing no faces found error.  

Edit: I installed all plug-ins and it worked

I'm not sure what your problem is
I made a short video and posted it in the original topic
Please refer to that

//deep.whitecatchel.ru/literotica/forums/thread-mve-machine-video-editor-guide?pid=31022#pid31022

I manage to upload aligned folder to frame folder somehow. but I cound't find the Roy B batch file in the tutorial video.

---

Why does it keep saying "no faces found"?  I have executed "5) data_dst faceset extract". and I got "aligned" and  "aligned_debug" folder. and then, I put alined,aligned_debug, frame folder into MVE just like tutorial says but it showing no faces found error.  

Edit: I installed all plug-ins and it worked

I'm not sure what your problem is
I made a short video and posted it in the original topic
Please refer to that

//deep.whitecatchel.ru/literotica/forums/thread-mve-machine-video-editor-guide?pid=31022#pid31022

Thanks to you, I manage to upload aligned folder to frame folder somehow. but I cound't find the Roy B batch file in the tutorial video.

 

[SPT]

: tried your script, either there's some problem coming from me or still some issue with the script : I had to delete manually the already corrected by Manual Re-extract frames from the MVE project\data_dst\aligned folder (which is easy because it is square aligned pics and the rest is widescreen full frames from source dst) and then I'm left with the dst full frames I couldn't correct with Manual re-extract in MVE project\data_dst\aligned, which is what I wanted.
But then in FRAMES folder, it's all full frames from dst, corresponding to all the frames deleted from aligned_debug, including successfully re-extracted frames with MANUAL re-extract + failed ones.

Which is basically useless for I want to do, but what I want to do might not even be possible in MVE, and if so, my idea is not that usefull.

Let me explain : I want to re-extract from full dst frames in MVE, not just correct wrong already aligned frames. So why would I need the FRAMES folder ? If I redo everything manually from full frames in MVE ?
But also, since I've never done it (I waited for your script to exist) maybe it's not even possible ?

Anyways, there's something I still don't understand.

: do you already have a tutorial covering what I want to do with MVE ? Is it even possible ?

: another question : I only have "aligned" and "FRAMES" folders in MVE project\data_dst
I thought there was supposed to be 3 folders at some point but I'm not sure. Or was it always 2 folders ?

 

[siao]

Why does it keep saying "no faces found"?  I have executed "5) data_dst faceset extract". and I got "aligned" and  "aligned_debug" folder. and then, I put alined,aligned_debug, frame folder into MVE just like tutorial says but it showing no faces found error.  

Edit: I installed all plug-ins and it worked

I'm not sure what your problem is
I made a short video and posted it in the original topic
Please refer to that

//deep.whitecatchel.ru/literotica/forums/thread-mve-machine-video-editor-guide?pid=31022#pid31022

I manage to upload aligned folder to frame folder somehow. but I cound't find the Roy B batch file in the tutorial video.

---

Why does it keep saying "no faces found"?  I have executed "5) data_dst faceset extract". and I got "aligned" and  "aligned_debug" folder. and then, I put alined,aligned_debug, frame folder into MVE just like tutorial says but it showing no faces found error.  

Edit: I installed all plug-ins and it worked

I'm not sure what your problem is
I made a short video and posted it in the original topic
Please refer to that

//deep.whitecatchel.ru/literotica/forums/thread-mve-machine-video-editor-guide?pid=31022#pid31022

Thanks to you, I manage to upload aligned folder to frame folder somehow. but I cound't find the Roy B batch file in the tutorial video.

---

You have to copy his code

Save as .bet

//deep.whitecatchel.ru/literotica/forums/thre...s-and-landmarks-re-extract?pid=30890#pid30890

 

[sbep]

"no faces found" for me too. also noticed it can't detect my GPU in settings, but not sure that matters in this case. i tried installing all the plugins (damn redundant, btw... each one installs the same few gb of packages from scratch rather than co-using..), but no luck.

Did you run this step?

Try it

that fixed it, thank you!!

 

[angeloshredder]

: tried your script, either there's some problem coming from me or still some issue with the script : I had to delete manually the already corrected by Manual Re-extract frames from the MVE project\data_dst\aligned folder (which is easy because it is square aligned pics and the rest is widescreen full frames from source dst) and then I'm left with the dst full frames I couldn't correct with Manual re-extract in MVE project\data_dst\aligned, which is what I wanted.
But then in FRAMES folder, it's all full frames from dst, corresponding to all the frames deleted from aligned_debug, including successfully re-extracted frames with MANUAL re-extract + failed ones.

Which is basically useless for I want to do, but what I want to do might not even be possible in MVE, and if so, my idea is not that usefull.

Let me explain : I want to re-extract from full dst frames in MVE, not just correct wrong already aligned frames. So why would I need the FRAMES folder ? If redo everything manually from full frames in MVE ?
But also, since I've never done it (I waited for your script to exist) maybe it's not even possible ?

Anyways, there's something I still don't understand.

: do you already have a tutorial covering what I want to do with MVE ? Is it even possible ?

: another question : I only have "aligned" and "FRAMES" folders in MVE project\data_dst
I thought there was supposed to be 3 folders at some point but I'm not sure. Or was it always 2 folders ?

:
" I want to re-extract from full dst frames in MVE, not just correct wrong already aligned frames. So why would I need the FRAMES folder ? If redo everything manually from full frames in MVE ? "
Yes , you can do it,  i created FRAMES folder just to make it easy to understand 
but in your case you have to open folder DST as the main project folder and change Detection Management / image information for this container  to VIDEO FRAMES
then in left menu select ALIGNED folder ( open images ) then change Detection Management / image information for this container  to FACE

if your DST does not contain ALIGNED and you want to extract faces from DST then  , You have to do 2 THINGS : DETECT FACES IN FRAMES and EXTRACT FACES TO ALIGNED !!!

1- select all frames (click on the first frame then CTRL+A , and  use the second menu ETECTIONS 
load both models : the native one + DLIB one

 

change the threshold until you get all faces in frames detected , when done you should get all frames with faces marked
then you can preview the faces and click in the faces you don't want to be cropped and right click to delete them , you can also click on the faces you want to keep and right click ( delete others ) to delete all others not needed.

when everything is done , then you can continue to the next step :
2-  Extracting and Cropping which is in the first menu Called Detection Management

Select USE DFL FACE EXTRACT

Start extracting

MVE has lot of sorting methodes to clean facesets , which are better and faster than DFL tools , they need a special tutorial to cover all of them

 

[SPT]

: tried your script, either there's some problem coming from me or still some issue with the script : I had to delete manually the already corrected by Manual Re-extract frames from the MVE project\data_dst\aligned folder (which is easy because it is square aligned pics and the rest is widescreen full frames from source dst) and then I'm left with the dst full frames I couldn't correct with Manual re-extract in MVE project\data_dst\aligned, which is what I wanted.
But then in FRAMES folder, it's all full frames from dst, corresponding to all the frames deleted from aligned_debug, including successfully re-extracted frames with MANUAL re-extract + failed ones.

Which is basically useless for I want to do, but what I want to do might not even be possible in MVE, and if so, my idea is not that usefull.

Let me explain : I want to re-extract from full dst frames in MVE, not just correct wrong already aligned frames. So why would I need the FRAMES folder ? If redo everything manually from full frames in MVE ?
But also, since I've never done it (I waited for your script to exist) maybe it's not even possible ?

Anyways, there's something I still don't understand.

: do you already have a tutorial covering what I want to do with MVE ? Is it even possible ?

: another question : I only have "aligned" and "FRAMES" folders in MVE project\data_dst
I thought there was supposed to be 3 folders at some point but I'm not sure. Or was it always 2 folders ?

:
" I want to re-extract from full dst frames in MVE, not just correct wrong already aligned frames. So why would I need the FRAMES folder ? If redo everything manually from full frames in MVE ? "
Yes , you can do it,  i created FRAMES folder just to make it easy to understand 
but in your case you have to open folder DST as the main project folder and change Detection Management / image information for this container  to VIDEO FRAMES
then in left menu select ALIGNED folder ( open images ) then change Detection Management / image information for this container  to FACE

if your DST does not contain ALIGNED and you want to extract faces from DST then  , You have to do 2 THINGS : DETECT FACES IN FRAMES and EXTRACT FACES TO ALIGNED !!!

1- select all frames (click on the first frame then CTRL+A , and  use the second menu ETECTIONS 
load both models : the native one + DLIB one

 

change the threshold until you get all faces in frames detected , when done you should get all frames with faces marked
then you can preview the faces and click in the faces you don't want to be cropped and right click to delete them , you can also click on the faces you want to keep and right click ( delete others ) to delete all others not needed.

when everything is done , then you can continue to the next step :
2-  Extracting and Cropping which is in the first menu Called Detection Management

Select USE DFL FACE EXTRACT

Start extracting

Thank you. But just to be clear, since I will use this for frames DFL (even with Manual Re-extract) couldn't extract, can I use MVE to manually re-extract (with the MVE system where you move brows, mouth, eyes...etc landmarks) and then save as DFL compatible aligned faces ?

 

[angeloshredder]

yes you can , for example you can copy a face landmarks from any frame and past it to the frame without face landmarks ( hard to detect) , then edit this pasted face manualy in the editor

 

[SPT]

yes you can , for example you can copy a face landmarks from any frame and past it to the frame without face landmarks ( hard to detect) , then edit this pasted face manualy in the editor

Thank you very much. Will try this back up solution next time I have too many undetected frames.

 

[Roy_B]

I cannot get DFDNET or PSFRGAN features to work within MVE. Anyone can explain how to do it?

I currently use  for DFDNET, and would like to understand if PSFRGAN works better.



i've tried to answer your questions in private (hope it fix your problems)

 

[SPT]

I cannot get DFDNET or PSFRGAN features to work within MVE. Anyone can explain how to do it?

I currently use  for DFDNET, and would like to understand if PSFRGAN works better.



i've tried to answer your questions in private (hope it fix your problems)

Never tried DFDNET, but all I know is there is additional stuff to install for it in MVE (it is not installed at the beginning, like the other dependencies) has it installed in MVE, he should know more about it. (But will probably tell you DFDNET is shit just use REMINI )

Seen your message. Thank you. I think maybe I found what I did wrong. Will try it when I have time and let you know.

 

[angeloshredder]

it seems like DFDNet installer has problems dowloading pretrained models ? So you need to download them manually from
When done , you need to put them in sub-folders of this folder : 
.\MachineEditor path      .\resources\external\plugins\DFDNet\project
but keep this structure : if a sub-folder doesn't existe you need to create it manually !!

after that test it only on faces ( not frames ) 

Warning : you can also try to use this more recent DLIB pretrained models :   .. but it's risky : you need to make some changes in their names to match the original ones and put them in the right directories : .\resources\external\plugins\DFDNet\project\packages

 

[siao]

: tried your script, either there's some problem coming from me or still some issue with the script : I had to delete manually the already corrected by Manual Re-extract frames from the MVE project\data_dst\aligned folder (which is easy because it is square aligned pics and the rest is widescreen full frames from source dst) and then I'm left with the dst full frames I couldn't correct with Manual re-extract in MVE project\data_dst\aligned, which is what I wanted.
But then in FRAMES folder, it's all full frames from dst, corresponding to all the frames deleted from aligned_debug, including successfully re-extracted frames with MANUAL re-extract + failed ones.


-----------
When you finish the manual extraction in DFL
Then you have to clear the recycle bin
And re-delete the uncorrected alignment pictures
Then run Roy_B's script
So you can get the picture you want
-----------


: another question : I only have "aligned" and "FRAMES" folders in MVE project\data_dst
I thought there was supposed to be 3 folders at some point but I'm not sure. Or was it always 2 folders ?


----------
I think most of "MANUAL RE-EXTRACT DELETED ALIGNED_DEBUG"
work.
DFL is faster
Search and fix bad alignment pictures in MVE
Not really easy to use
When i think so
Then
's script appears
Thank him very much

But some DFL can’t handle
Maybe it can
-------

Personal feeling 
for reference only

 

[Roy_B]

Thanks for help, manual install works both for DFDNET and PSFRGAN. I've tested both and imho DFDNET it's still a better choice for DFL (PSFRGAN is less accurate i think).

The only problem now is that if i use DFDNET into MVE it gives me this "GPU out of memory" error:
%ERROR%$Error in enhancing this image: CUDA out of memory. Tried to allocate 758.00 MiB (GPU 0; 6.00 GiB total capacity; 4.34 GiB already allocated; 84.50 MiB free; 4.68 GiB reserved in total by PyTorch)

Apparently my GTX 1060 6GB is not enough to handle it... But i have no problems since i use Colab version for that.



Regarding the "third folder" (MVE Project\data_dst\RE-EXTRACTED) i preferred to let the user decide, but it's easy to include into the script if needed:

just add these 2 entries, respectively just below the lines ":SOURCE" and "ESTINATION"
MD "MVE Project\data_src\RE-EXTRACTED"
MD "MVE Project\data_dst\RE-EXTRACTED"

 

[angeloshredder]

Thanks for help, manual install works both for DFDNET and PSFRGAN. I've tested both and imho DFDNET it's still a better choice for DFL (PSFRGAN is less accurate i think).

The only problem now is that if i use DFDNET into MVE it gives me this "GPU out of memory" error:
%ERROR%$Error in enhancing this image: CUDA out of memory. Tried to allocate 758.00 MiB (GPU 0; 6.00 GiB total capacity; 4.34 GiB already allocated; 84.50 MiB free; 4.68 GiB reserved in total by PyTorch)

Apparently my GTX 1060 6GB is not enough to handle it... But i have no problems since i use Colab version for that.



Regarding the "third folder" (MVE Project\data_dst\RE-EXTRACTED) i preferred to let the user decide, but it's easy to include into the script if needed:

just add these 2 entries, respectively just below the lines ":SOURCE" and "ESTINATION"
MD "MVE Project\data_src\RE-EXTRACTED"
MD "MVE Project\data_dst\RE-EXTRACTED"

"PSFRGAN is less accurate i think " , i agree .. i don't like it

 

[kakks]

DFL face data_dst extraction is pretty awful when angle is showing only partial part of the face. I faced many face landmarks are broken when extracting result of a blowjob video. then I have to "edit alignment" using MVE hundreds of files one by one. MVE is pretty useful so far but Is there any way to fix a lot of broken landmarks easier?

 

[SPT]

DFL face data_dst extraction is pretty awful when angle is showing only partial part of the face. I faced many face landmarks are broken when extracting result of a blowjob video. then I have to "edit alignment" using MVE hundreds  of files one by one. MVE is pretty useful so far but Is there any way to fix a lot of broken landmarks  easier?

If you already used DFL Manual re-extract, and then MVE for the pics Manual re-extract didn't save, there's nothing else but just pre-edit video cutting the parts you know won't work before hand. You can do that with any video editor.

 

[kakks]

I imported faces from aligned to Frame folder and click save button. but if i re-open the Frame folder, the faces i imported are gone. Do I have to import hundreds of faces everytime i need to modificate faces on Frame folder?

 

[siao]

I imported faces from aligned to Frame folder and click save button. but if i re-open the Frame folder, the faces i imported are gone. Do I have to import hundreds of faces everytime i need to modificate faces on Frame folder?

This one?

 

[boonino]

I have a question about the face information after edits.  I've fixed up a few frames that have extreme pitch/roll/yaws.  I've noticed the edits don't update these values which I believe the set creator uses.  Is there a way to do this?

 

[Agamemnus]

[font=Roboto, Arial, sans-serif]MVE / REMINI UTILS : PACK - REMINI - UNPACK  .... TO ENHANCE YOUR  FACESTS[/font]

This my second MVE tutorial : 

samples from faceset enhaced using MVE technique : pack-remini-unpack Compared with original ones side by side :

sample of enhanced packed picture

---

CAUTION : 
When you import embedded data , you need to save the changes by clicking on the top of every opened tab ( faces , frames .. )  , then close the tabs , refresh project folders then open folders again ..
if your faces and frames don't show green boxes then importation did not happen , you need to re-import data ..

---

Great tutorial, love it.