calcuis

/

pig-vae

like 24

you could refer to gguf-node (pypi|repo|pack); check convertor zero; it should be able to convert any safetensors to gguf

Is the qwen_image VAE the same as Wan2.1 vae?
I notice that this gguf is excact the same size as comfy´s qwen VAE . Qwens vae is the same as wan2.1 vae.

but the hash is different; you could test them both with wan video and qwen image models; see can use it interchange and give the same output or not

most of the time, check the hash might give you the answer; for example, pig_flux_vae_fp32-f16.gguf and pig_1k_lumina_vae_fp32-f16.gguf share the same hash, these two can confirm exactly the same
another example, ltxv-13b-0.9.7-dev-fp32-q4_0.gguf and ltxv-13b-0.9.7-distilled-fp32-q4_0.gguf are same size but they are different, no matter the hash or the output behavior

does the converter zero just convert to gguf without quantizing ? if no quantizing is done what the benefits of a gguf?

you could make it with quantizor or cutter