pose-with-style icon indicating copy to clipboard operation
pose-with-style copied to clipboard

ask for advice

Open zcdliuwei opened this issue 2 years ago • 6 comments

Wonderful work! Can replace the source image with an animation character, and then use target pose to generate animation character that meet the target pose?

zcdliuwei avatar Mar 17 '23 11:03 zcdliuwei

You can but the results will probably not be great since the model was trained on the DeepFashion dataset.

On Fri, Mar 17, 2023 at 7:12 AM Keep Growing And Moving Forward < @.***> wrote:

Wonderful work! Can replace the source image with an animation character, and then use target pose to generate animation character that meet the target pose?

— Reply to this email directly, view it on GitHub https://github.com/BadourAlBahar/pose-with-style/issues/28, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGNJLVOFB6P3DJYOYK5JFJDW4RBKPANCNFSM6AAAAAAV6MSONI . You are receiving this because you are subscribed to this thread.Message ID: @.***>

BadourAlBahar avatar Mar 17 '23 12:03 BadourAlBahar

Body

Thank you for your reply

If I collect some images of anime characters and use your script to train them, will the effect be better ? like in real human scenes

zcdliuwei avatar Mar 17 '23 12:03 zcdliuwei

The methods builds on dense pose. If dense pose works well with the subjects you have and you have enough data to train on, I think it should work.

On Fri, Mar 17, 2023 at 8:58 AM Keep Growing And Moving Forward < @.***> wrote:

You can but the results will probably not be great since the model was trained on the DeepFashion dataset. … <#m_5188481289229552592_> On Fri, Mar 17, 2023 at 7:12 AM Keep Growing And Moving Forward < @.> wrote: Wonderful work! Can replace the source image with an animation character, and then use target pose to generate animation character that meet the target pose? — Reply to this email directly, view it on GitHub <#28 https://github.com/BadourAlBahar/pose-with-style/issues/28>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGNJLVOFB6P3DJYOYK5JFJDW4RBKPANCNFSM6AAAAAAV6MSONI https://github.com/notifications/unsubscribe-auth/AGNJLVOFB6P3DJYOYK5JFJDW4RBKPANCNFSM6AAAAAAV6MSONI . You are receiving this because you are subscribed to this thread.Message ID: @.>

Thank you for your reply

If I collect some images of anime characters and use your script to train them, will the effect be better ? like in real human scenes

— Reply to this email directly, view it on GitHub https://github.com/BadourAlBahar/pose-with-style/issues/28#issuecomment-1473803484, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGNJLVMGW53NLC2S4COVP5TW4RNVRANCNFSM6AAAAAAV6MSONI . You are receiving this because you commented.Message ID: @.***>

BadourAlBahar avatar Mar 17 '23 13:03 BadourAlBahar

thanks, i will give it a try !

zcdliuwei avatar Mar 17 '23 16:03 zcdliuwei

Hello, I use the same environment and inference script as you: python inference.py --input_path ./data --CCM_pretrained_model path/to/CCM_epoch50.pt --pretrained_model path/to/posewithstyle.pt

But the output I get is this: fashionWOMENDressesid0000262902_3back_2_fashionWOMENDressesid0000262902_1front_vis

Did I miss something

zcdliuwei avatar Mar 21 '23 03:03 zcdliuwei

Hello, I use the same environment and inference script as you: python inference.py --input_path ./data --CCM_pretrained_model path/to/CCM_epoch50.pt --pretrained_model path/to/posewithstyle.pt

But the output I get is this(View attachments) Did I miss something ?

At 2023-03-17 21:07:35, "Badour AlBahar" @.***> wrote:

The methods builds on dense pose. If dense pose works well with the subjects you have and you have enough data to train on, I think it should work.

On Fri, Mar 17, 2023 at 8:58 AM Keep Growing And Moving Forward < @.***> wrote:

You can but the results will probably not be great since the model was trained on the DeepFashion dataset. … <#m_5188481289229552592_> On Fri, Mar 17, 2023 at 7:12 AM Keep Growing And Moving Forward < @.> wrote: Wonderful work! Can replace the source image with an animation character, and then use target pose to generate animation character that meet the target pose? — Reply to this email directly, view it on GitHub <#28 https://github.com/BadourAlBahar/pose-with-style/issues/28>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGNJLVOFB6P3DJYOYK5JFJDW4RBKPANCNFSM6AAAAAAV6MSONI https://github.com/notifications/unsubscribe-auth/AGNJLVOFB6P3DJYOYK5JFJDW4RBKPANCNFSM6AAAAAAV6MSONI . You are receiving this because you are subscribed to this thread.Message ID: @.>

Thank you for your reply

If I collect some images of anime characters and use your script to train them, will the effect be better ? like in real human scenes

— Reply to this email directly, view it on GitHub https://github.com/BadourAlBahar/pose-with-style/issues/28#issuecomment-1473803484, or unsubscribe https://github.com/notifications/unsubscribe-auth/AGNJLVMGW53NLC2S4COVP5TW4RNVRANCNFSM6AAAAAAV6MSONI . You are receiving this because you commented.Message ID: @.***>

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

zcdliuwei avatar Mar 21 '23 08:03 zcdliuwei