drag gan on custom images | google colab | drag your gan

[@farithadnan]

Thank you, sir! I spent a couple of days setting up both DragGAN and PTI on my computer. I got stuck with inverted images with the .pt extension. I didn't know how to convert them to .pkl until I found your videos! 🔥

[@mattfarmerai]

Great job on this video. You are a legend that deserves more followers, keep up the great content

[@programming_hut]

♥️ I don’t need much followers I just need such sweet words to continue. ♥️

[@qwes420]

@@programming_hut can you help me please? iam using your notebook with custom images but i want to train on 512 not 1024 because its too slow and lagy, i've tried replacing StyleGAN on FFHQ 1024x1024 with my 512 ffhq but then a get an error at step 5 (AssertionError: ) can you edit the code? :))))

[@KingPorkkanen]

Not working anymore. It's giving these errors. First on optional step: sh: 0: cannot open scripts/download_model.sh: No such file Then when trying to run the last step. File "/content/DragGAN/visualizer_drag_gradio.py", line 72 if isinstance(global_state, gr.State): IndentationError: unexpected indent

[@GDj-x8y]

I'm having problems with files. The file "download_model.sh", supposedly it does not exist, only has a "download_model.py" (in the Scripts folder), so I can't download the models and complete the later steps. can you help me?

[@AI_FOR_VFX]

love it !,thank you very much !

[@okopyl]

Thanks. Could you please write in general what steps are required to use Drag your GAN on your own image?

[@AI_FOR_VFX]

How to improve the invert image to be more similar to the input image?

[@jigarshah8783]

in the optional step there is no sh file so it is giving an error in the code

[@chetancc]

Hi Brother, I found below files got created from "PTI_custom_model.ipynb" model_VRVOFYDEEUMF.pkl model_VRVOFYDEEUMF_personal_image.pt So I renamed "model_VRVOFYDEEUMF.pkl" to "stylegan2_custom_512_pytorch.pkl" and tried to use it in "draggan_custom_images.ipynb" but it didn't work in the end. Can you please check once again?

[@anonymousfz007]

How do I use custom dimesions , say a picture in potrait mode?

[@mohammad815]

Hi programming_hut , sorry i dont know your real name Your tutorial is fantastic! Could you share tips on using it with 'StyleGan_human,' especially in terms of pre-processing for human images? I'd greatly appreciate your insights. Also, I'm curious about your thoughts on 'insetGAN.' Thanks for your great work!

[@GameofSen]

hi is this only for faces? or can I use it lets say for cars or the fashion designs like the examples in the Drag Gan website?

[@junyanwu9328]

Hi, I followed your previous process, but he did not generate the pki file Stylegan2 Custom 512, what is the reason?

[@SRBstyles]

@programmingHut....sir..can you tell me...which version of python and opencv you used in your previous video...smart cctv camera

[@wanderlusttom]

Very cool

[@chalukyareddy6527]

local variable 'shape' referenced before assignment. I am getting this error when inverting image only for lips? Can someone help me in this?

[@okopyl]

Do you have to train the model with your own image so it can generate that image for dragging?

[@dongyunkim1148]

Thanks for the tutorial. I want to load my custom trained stylegan2-ada pkl file in DragGAN, but it is not loaded. Do you know what should I do to load it? I change the .pkl name correctly, e.g) stylegan2_STH.pkl, but it won't work...

[@spinninglink]

How do you upload custom full human pictures? This one seems to only crop to the face

[@programming_hut]

In pti instead of ffhq try it with full body human model

[@spinninglink]

@@programming_hut I mean it happens in Step 4 - preprocess data. It only crops out the face of the final image. It doesn't keep the rest of it.

[@enatdoroshev8162]

Did you find a way to do a full body or still not working ?

[@spinninglink]

@@enatdoroshev8162 nah, but i didn't mess around with it much. The tool itself is not that good, it deforms and makes people unrecognizable after just 1 drag.. not worth it for now imo

[@SilentPraniii]

❤

[@homiraghuvanshi5867]

How to test draggan on other custom images (not human faces/humans)

[@ErickXu-ez9hq]

[Errno 2] No such file or directory: '/content/PTI/image_processed/image.jpg'

[@ErickXu-ez9hq]

I know the reason for this problem, pre_process_images(f'/content/PTI/{image_dir_name}_original'), this function is to segment the image according to the range of the face, if the uploaded image does not contain a face or is blocked The local variable 'shape' referenced before assignment error was reported because the face could not be recognized, resulting in no corresponding processed image being generated, so aligned_image = Image.open(f'/content/PTI/{image_dir_name}_processed/{image_name} afterwards .jpeg') This statement will report an error that the corresponding picture cannot be found

[@deepabala4232]

Yes I am also facing the same issue. Can anyone share what type of image can we use to get better results.?

[@ErickXu-ez9hq]

@@deepabala4232 unblocked face image

[@rahathasan819]

Can this be done locally using my own GPU?

[@patrikhollender3232]

Do you only get an altered image? Or can you get a mp4 with the alteration?

[@patrikhollender3232]

NO?

[@LUXSON_Tech_Demos]

FileNotFoundError: [Errno 2] No such file or directory: '/content/PTI/image_processed/image.jpeg' I processed the image but it doesnt appear in the folder

[@shyampatel7755]

facing same issue let me know if you find any solution

[@programming_hut]

Try again running all cells properly and put original image in original folder

[@pabloimpablo85]

Same issue here. Have you found any solution?

[@pabloimpablo85]

I was able to solve the issue. You have to replace 'image_name' with the name of your image ( in this part of the code: original_image = Image.open(f'{image_name}.jpg'). In my case using the name "image" triggered the problem, so use any other =) . You'll have to do this two or three times during the process in different cells.

[@pabloimpablo85]

@@shyampatel7755 Try doing what I detailed in the other comment, hope that helps =)

[@nehacool28]

I get this error inside run_PTI assert target.shape == (G.img_channels, G.img_resolution, G.img_resolution)

[@theAi_journey]

how did you solve it?

[@KalkiKrivaDNA]

Can we use drag gan on mobile?

[@preenxus3425]

show more please!!! turn her look around, open/close her mouth, more demonstration please!

[@Vishnu-9-30p]

Please confusion this video .tell me new relased .and step by step slowly please

[@leareviron3159]

the firts collab notebook seems to have been deleted ://

[@programming_hut]

Give me some time, deleted accidentally.

[@leareviron3159]

@@programming_hut thanks !

[@programming_hut]

hi, thanks for letting me know, I have updated the colab 1: colab.research.google.com/drive/1aZW21clDPWLvwBnaQG5Xcpc_U-8zy3ow?usp=sharing

[@richardb8317]

too much work