Spaces:
Runtime error
Runtime error
Update upscaler_specialist.py
Browse files- upscaler_specialist.py +22 -22
upscaler_specialist.py
CHANGED
|
@@ -1,31 +1,32 @@
|
|
| 1 |
# upscaler_specialist.py
|
| 2 |
-
# Copyright (C) 2025 Carlos Rodrigues
|
| 3 |
# Especialista ADUC para upscaling espacial de tensores latentes.
|
| 4 |
|
| 5 |
import torch
|
| 6 |
import logging
|
| 7 |
from diffusers import LTXLatentUpsamplePipeline
|
| 8 |
-
|
| 9 |
from ltx_manager_helpers import ltx_manager_singleton
|
| 10 |
|
| 11 |
logger = logging.getLogger(__name__)
|
| 12 |
|
| 13 |
-
|
| 14 |
class UpscalerSpecialist:
|
| 15 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
self.device = "cuda" if torch.cuda.is_available() else "cpu"
|
| 17 |
-
self.cpu_device = torch.device("cpu")
|
| 18 |
-
self.workspace_dir = workspace_dir
|
| 19 |
-
self.pipe_upsample = None
|
| 20 |
self.base_vae = None
|
|
|
|
| 21 |
|
| 22 |
def _lazy_init(self):
|
| 23 |
-
"""Inicializa
|
| 24 |
if self.base_vae is None:
|
| 25 |
try:
|
| 26 |
-
from ltx_manager_helpers import ltx_manager_singleton
|
| 27 |
if ltx_manager_singleton.workers:
|
| 28 |
self.base_vae = ltx_manager_singleton.workers[0].pipeline.vae
|
|
|
|
| 29 |
else:
|
| 30 |
logger.warning("[Upscaler] Nenhum worker disponível no ltx_manager_singleton.")
|
| 31 |
except Exception as e:
|
|
@@ -34,7 +35,6 @@ class UpscalerSpecialist:
|
|
| 34 |
|
| 35 |
if self.pipe_upsample is None and self.base_vae is not None:
|
| 36 |
try:
|
| 37 |
-
from ltx_video.pipelines.latent_upscale import LTXLatentUpsamplePipeline
|
| 38 |
self.pipe_upsample = LTXLatentUpsamplePipeline.from_pretrained(
|
| 39 |
"linoyts/LTX-Video-spatial-upscaler-0.9.8",
|
| 40 |
vae=self.base_vae,
|
|
@@ -44,25 +44,25 @@ class UpscalerSpecialist:
|
|
| 44 |
except Exception as e:
|
| 45 |
logger.error(f"[Upscaler] Falha ao carregar pipeline: {e}")
|
| 46 |
|
|
|
|
| 47 |
def upscale(self, latents: torch.Tensor) -> torch.Tensor:
|
|
|
|
| 48 |
self._lazy_init()
|
| 49 |
if self.pipe_upsample is None:
|
| 50 |
logger.warning("[Upscaler] Pipeline indisponível. Retornando latentes originais.")
|
| 51 |
return latents
|
|
|
|
| 52 |
try:
|
| 53 |
-
|
| 54 |
-
|
|
|
|
| 55 |
return result.latents
|
| 56 |
except Exception as e:
|
| 57 |
-
logger.error(f"[Upscaler] Erro durante upscale: {e}")
|
| 58 |
return latents
|
| 59 |
-
|
| 60 |
|
| 61 |
-
|
| 62 |
-
#
|
| 63 |
-
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
except Exception as e:
|
| 67 |
-
logger.error(f"Não foi possível inicializar o UpscalerSpecialist Singleton: {e}")
|
| 68 |
-
upscaler_specialist_singleton = None
|
|
|
|
| 1 |
# upscaler_specialist.py
|
| 2 |
+
# Copyright (C) 2025 Carlos Rodrigues
|
| 3 |
# Especialista ADUC para upscaling espacial de tensores latentes.
|
| 4 |
|
| 5 |
import torch
|
| 6 |
import logging
|
| 7 |
from diffusers import LTXLatentUpsamplePipeline
|
|
|
|
| 8 |
from ltx_manager_helpers import ltx_manager_singleton
|
| 9 |
|
| 10 |
logger = logging.getLogger(__name__)
|
| 11 |
|
|
|
|
| 12 |
class UpscalerSpecialist:
|
| 13 |
+
"""
|
| 14 |
+
Especialista responsável por aumentar a resolução espacial de tensores latentes
|
| 15 |
+
usando o LTX Video Spatial Upscaler.
|
| 16 |
+
"""
|
| 17 |
+
def __init__(self):
|
| 18 |
+
# Força uso de CUDA se disponível
|
| 19 |
self.device = "cuda" if torch.cuda.is_available() else "cpu"
|
|
|
|
|
|
|
|
|
|
| 20 |
self.base_vae = None
|
| 21 |
+
self.pipe_upsample = None
|
| 22 |
|
| 23 |
def _lazy_init(self):
|
| 24 |
+
"""Inicializa VAE e pipeline apenas quando necessário."""
|
| 25 |
if self.base_vae is None:
|
| 26 |
try:
|
|
|
|
| 27 |
if ltx_manager_singleton.workers:
|
| 28 |
self.base_vae = ltx_manager_singleton.workers[0].pipeline.vae
|
| 29 |
+
logger.info("[Upscaler] VAE base obtido com sucesso.")
|
| 30 |
else:
|
| 31 |
logger.warning("[Upscaler] Nenhum worker disponível no ltx_manager_singleton.")
|
| 32 |
except Exception as e:
|
|
|
|
| 35 |
|
| 36 |
if self.pipe_upsample is None and self.base_vae is not None:
|
| 37 |
try:
|
|
|
|
| 38 |
self.pipe_upsample = LTXLatentUpsamplePipeline.from_pretrained(
|
| 39 |
"linoyts/LTX-Video-spatial-upscaler-0.9.8",
|
| 40 |
vae=self.base_vae,
|
|
|
|
| 44 |
except Exception as e:
|
| 45 |
logger.error(f"[Upscaler] Falha ao carregar pipeline: {e}")
|
| 46 |
|
| 47 |
+
@torch.no_grad()
|
| 48 |
def upscale(self, latents: torch.Tensor) -> torch.Tensor:
|
| 49 |
+
"""Aplica o upscaling 2x nos tensores latentes fornecidos."""
|
| 50 |
self._lazy_init()
|
| 51 |
if self.pipe_upsample is None:
|
| 52 |
logger.warning("[Upscaler] Pipeline indisponível. Retornando latentes originais.")
|
| 53 |
return latents
|
| 54 |
+
|
| 55 |
try:
|
| 56 |
+
logger.info(f"[Upscaler] Recebido shape {latents.shape}. Executando upscale em {self.device}...")
|
| 57 |
+
result = self.pipe_upsample(latents=latents, output_type="latent")
|
| 58 |
+
logger.info(f"[Upscaler] Upscale concluído. Novo shape: {result.latents.shape}")
|
| 59 |
return result.latents
|
| 60 |
except Exception as e:
|
| 61 |
+
logger.error(f"[Upscaler] Erro durante upscale: {e}", exc_info=True)
|
| 62 |
return latents
|
|
|
|
| 63 |
|
| 64 |
+
|
| 65 |
+
# ---------------------------
|
| 66 |
+
# Singleton global
|
| 67 |
+
# ---------------------------
|
| 68 |
+
upscaler_specialist_singleton = UpscalerSpecialist()
|
|
|
|
|
|
|
|
|