feat: trained Minimax VAE Latent2rgb proj (#1856)

This commit is contained in:
stduhpf 2026-08-05 19:39:10 +02:00 committed by GitHub
parent ea7f0c87cf
commit bfbef5b7e6
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 35 additions and 0 deletions

View File

@ -4,6 +4,33 @@
#include "core/tensor.hpp" #include "core/tensor.hpp"
#include "ggml.h" #include "ggml.h"
const float minimax_latent_rgb_proj[24][3] = {
{0.19819857f, 0.11584999f, 0.07929777f},
{-0.16047224f, -0.10601170f, -0.15996324f},
{0.47391951f, 0.37602475f, 0.20267826f},
{-0.09857441f, -0.27435449f, -0.51681751f},
{-0.18930605f, -0.10512278f, -0.28571478f},
{-0.15639569f, -0.18000929f, -0.25432852f},
{-0.07176921f, -0.10901598f, -0.06654253f},
{-0.05014077f, -0.05839826f, -0.05516087f},
{-0.05201424f, -0.04351913f, -0.01507579f},
{0.24750438f, 0.13307422f, 0.17684120f},
{0.07377446f, 0.10235858f, 0.11707827f},
{0.02908304f, 0.06587022f, 0.10643690f},
{-0.00670531f, -0.03857879f, 0.01750151f},
{-0.07119107f, -0.03083323f, -0.01995450f},
{-0.08612627f, -0.07253841f, -0.01442890f},
{0.08793202f, 0.08681750f, 0.02994647f},
{0.00876893f, 0.02721868f, 0.00091178f},
{-0.03484412f, -0.02711262f, -0.00110101f},
{-0.00679772f, -0.01844275f, -0.01683359f},
{0.04287028f, 0.01601068f, 0.04037397f},
{-0.00493432f, -0.00230528f, 0.00353911f},
{0.01495088f, 0.00292306f, 0.00416671f},
{0.00495307f, 0.05066542f, 0.05210543f},
{-0.02154842f, -0.01518524f, 0.00442402f}};
float minimax_latent_rgb_bias[3] = {0.07776964f, -0.01580954f, -0.06561434f};
const float ltxav_latent_rgb_proj[128][3] = { const float ltxav_latent_rgb_proj[128][3] = {
{-0.0293802f, -0.0362516f, -0.0291386f}, {-0.0293802f, -0.0362516f, -0.0291386f},
{0.0117735f, 0.0223435f, 0.018856f}, {0.0117735f, 0.0223435f, 0.018856f},

View File

@ -2296,6 +2296,14 @@ public:
LOG_WARN("No latent to RGB projection known for this model"); LOG_WARN("No latent to RGB projection known for this model");
return; return;
} }
} else if (channels == 24) {
if(sd_version_is_minimax_h3(version)){
latent_rgb_proj = minimax_latent_rgb_proj;
latent_rgb_bias = minimax_latent_rgb_bias;
} else {
LOG_WARN("No latent to RGB projection known for this model");
return;
}
} else if (channels == 16) { } else if (channels == 16) {
if (sd_version_is_sd3(version)) { if (sd_version_is_sd3(version)) {
latent_rgb_proj = sd3_latent_rgb_proj; latent_rgb_proj = sd3_latent_rgb_proj;