image-to-image translation / conditional GANs
pix2pix is a Torch (Lua) implementation of conditional GANs for paired image-to-image translation, mapping an input image to a corresponding output image.
Solves the same problem with a different architecture or at a different layer. Expect to rewrite the integration.
when you need unpaired image-to-image translation without aligned training pairs.
Software that can generate photos from paintings, turn horses into zebras, perform style transfer, and more.
when you need diverse (multimodal) outputs from a single input under paired supervision.
Toward Multimodal Image-to-Image Translation
when you need unsupervised translation using a shared latent space assumption.
Unsupervised Image-to-Image Translation
when you need multimodal outputs in unsupervised image-to-image translation.
Multimodal Unsupervised Image-to-Image Translation
when you need very fast one-step image-to-image translation using diffusion models.
One-step image-to-image with Stable Diffusion turbo: sketch2image, day2night, and more
when you want a broad GAN library on PaddlePaddle with many image-to-image applications.
PaddlePaddle GAN library, including lots of interesting applications like First-Order motion transfer, Wav2Lip, picture repair, image editing, photo2cartoon, image style transfer, GPEN, and so on.
An earlier or more limited way to do the job. Still the right call when you need something small, proven, or CPU-only.
when you specifically need image deblurring and want a pretrained GAN model.
Image Deblurring using Generative Adversarial Networks
when you specifically need portrait-to-cartoon translation and don't need a general framework.
人像卡通化探索项目 (photo-to-cartoon translation project)
These projects were analysed and named pix2pix among their alternatives. The relationship is not symmetric — how pix2pix rates them is a separate judgement, made when pix2pix is analysed in its own right.
calls pix2pix “Choose for the canonical pix2pix model if you need to match a legacy Torch codebase and don't need other GAN applications.”
PaddlePaddle GAN library, including lots of interesting applications like First-Order motion transfer, Wav2Lip, picture repair, image editing, photo2cartoon, image style transfer, GPEN, and so on.