ComfyUI Texture Upscaling (Batch+transparency+filename) AIO

[JuanchoTexHD]

FINAL UPDATE
Finally completed workflows for flux and illustrious (ADDED LUMINA2 TOO)
LINK: look attachments below

 

[JuanchoTexHD]

Updated workflows: 3 upscaling methods

 

[TexMaster]

Hi sorry I wanted to ask a question Can you do a little tutorial on YouTube by any chance and I know and I read it in the description I tried to do a couple of tests but I can't do even one because I have a 3070ti and I have 16 GB too and I wanted to ask you if you could do a tutorial on it so I understand well what I should do how I should move I tried too but it's a bit too complicated to see various things or the nodes various things there and I didn't understand anything

 

[JuanchoTexHD]

Hi sorry I wanted to ask a question Can you do a little tutorial on YouTube by any chance and I know and I read it in the description I tried to do a couple of tests but I can't do even one because I have a 3070ti and I have 16 GB too and I wanted to ask you if you could do a tutorial on it so I understand well what I should do how I should move I tried too but it's a bit too complicated to see various things or the nodes various things there and I didn't understand anything

Check on top post, a simpler workflow on video, use english subs

 

[TexMaster]

AGGIORNAMENTO 2; ecco un flusso di lavoro più semplice, basta caricare il file json sull'interfaccia comoda e aggiornare i nodi
Collegamento al flusso di lavoro con spiegazione video (utilizza la traduzione automatica dei sottotitoli) nella descrizione

Grazie per il tutorial. Ora lo guardo tutto.
[unione automatica]1746268508[/unione automatica]

AGGIORNAMENTO 2; ecco un flusso di lavoro più semplice, basta caricare il file json sull'interfaccia comoda e aggiornare i nodi
Collegamento al flusso di lavoro con spiegazione video (utilizza la traduzione automatica dei sottotitoli) nella descrizione

Ciao, mi dispiace, non riesco a trovare il modello realismIllustriousBy_v35FP1 / real hat gan sharper

Post automatically merged: May 3, 2025

Hey!
I wanted to send you this workflow I found online that I really liked.
It's called Clarity AI — it's a model that creates pretty nice textures and gives them a more realistic touch.
I tested it out, and it works quite well, although there are still a few things that could be improved.


I’d like to share it with you, so maybe you can help refine it and make it even better.
Once it's cleaned up and fully optimized, we could also publish it on GBATEMP as a free resource for everyone.


Let me know what you think!

 

[JuanchoTexHD]

Last update: New workflows with complete batch process original filenames and transparency

 

[TexMaster]

Last update: New workflows with complete batch process original filenames and transparency

could you post the missing files i cant find them on internet

 

[JuanchoTexHD]

could you post the missing files i cant find them on internet

Sure, this flux model WORKS WELL ON 7 STEPS not 4, my fault on video
FLUX HEAVY: h ttps://civitai.com/models/1119082?modelVersionId=1258538

Q6_K Version​

• VRAM Requirement: 16GB
• Best balance of quality and performance
• Recommended for users with RTX 3080/3090 or similar GPUs

Q5_K_M Version​

• VRAM Requirement: 12GB
• Good quality with reduced memory footprint
• Ideal for RTX 3060 Ti/3070/2080 Ti users

Q3_K_L Version​

• VRAM Requirement: 8GB
• Most memory-efficient version
• Enables running on mid-range GPUs like RTX 3060/2060 Super

T5 ENCODER : h ttps://huggingface.co/city96/t5-v1_1-xxl-encoder-gguf/tree/main
CLIP L : h ttps://huggingface.co/zer0int/CLIP-GmP-ViT-L-14/blob/main/ViT-L-14-TEXT-detail-improved-hiT-GmP-TE-only-HF.safetensors
VAE ENCODER: https://huggingface.co/ffxvs/vae-flux/blob/main/ae.safetensors
-----------------------------------------------------------------------------------------------------------------------------------
ILLUSTRIOUS REALISTIC: h ttps://civitai.com/models/974693?modelVersionId=1736657
all sizes are the same 6.46GB 20 steps
--------------------------------------------------------------------------------------------------------------------------------------
For the esrgan upscalers look on openmodelDB there is a lot of upscalers
-----------------------------------------------------------------------------------------------------------------------------------------
ANOTHER MODEL lighter than flux and better than XL, not fast as illustrious but faster than flux

LUMINA 2 MODELS: h ttps://huggingface.co/calcuis/lumina-gguf/tree/main
TEXT ENCODER: h ttps://huggingface.co/calcuis/lumina-gguf/blob/main/gemma_2_2b_fp16.safetensors
VAE: h ttps://huggingface.co/calcuis/lumina-gguf/blob/main/lumina2_vae_fp8.safetensors

The 3 workflows are attached below

 

[TexMaster]

Sure, this flux model WORKS WELL ON 7 STEPS not 4, my fault on video
FLUX HEAVY: h ttps://civitai.com/models/1119082?modelVersionId=1258538

Q6_K Version​

• VRAM Requirement: 16GB
• Best balance of quality and performance
• Recommended for users with RTX 3080/3090 or similar GPUs

Q5_K_M Version​

• VRAM Requirement: 12GB
• Good quality with reduced memory footprint
• Ideal for RTX 3060 Ti/3070/2080 Ti users

Q3_K_L Version​

• VRAM Requirement: 8GB
• Most memory-efficient version
• Enables running on mid-range GPUs like RTX 3060/2060 Super

T5 ENCODER : h ttps://huggingface.co/city96/t5-v1_1-xxl-encoder-gguf/tree/main
CLIP L : h ttps://huggingface.co/zer0int/CLIP-GmP-ViT-L-14/blob/main/ViT-L-14-TEXT-detail-improved-hiT-GmP-TE-only-HF.safetensors
VAE ENCODER: https://huggingface.co/ffxvs/vae-flux/blob/main/ae.safetensors
-----------------------------------------------------------------------------------------------------------------------------------
ILLUSTRIOUS REALISTIC: h ttps://civitai.com/models/974693?modelVersionId=1736657
all sizes are the same 6.46GB 20 steps
--------------------------------------------------------------------------------------------------------------------------------------
For the esrgan upscalers look on openmodelDB there is a lot of upscalers
-----------------------------------------------------------------------------------------------------------------------------------------
ANOTHER MODEL lighter than flux and better than XL, not fast as illustrious but faster than flux

LUMINA 2 MODELS: h ttps://huggingface.co/calcuis/lumina-gguf/tree/main
TEXT ENCODER: h ttps://huggingface.co/calcuis/lumina-gguf/blob/main/gemma_2_2b_fp16.safetensors
VAE: h ttps://huggingface.co/calcuis/lumina-gguf/blob/main/lumina2_vae_fp8.safetensors

The 3 workflows are attached below

Thank you

 

[TexMaster]

FINAL UPDATE
Finally completed workflows for flux and illustrious (ADDED LUMINA2 TOO)
LINK: look attachments below

FINAL UPDATE
Finally completed workflows for flux and illustrious (ADDED LUMINA2 TOO)
LINK: look attachments below

Ciao JuanchoTexHD!

IT
Ho provato tutti e tre i modelli che mi hai consigliato. Ti dico sinceramente che il risultato non mi ha convinto del tutto, ma te lo dico con il massimo rispetto: non è assolutamente una critica negativa. Anzi, apprezzo davvero tanto quello che fai e ti ringrazio di cuore, perché i modelli sono molto belli!


Quello che sto cercando però è un effetto più realistico, nello stile di un'immagine che ti vorrei mostrare qui in privato.
Se riesci ad aiutarmi a ottenere qualcosa di simile, te ne sarei davvero grato!


Se per farlo serve del tempo o un lavoro extra, fammi sapere anche se c’è un compenso: ne possiamo parlare tranquillamente in privato.
Ti dico già che non sono un esperto di ComfyUI, sto ancora imparando. Ho una RTX 3070 Ti, 16 GB di RAM e 8 GB di VRAM, quindi non so se riesco a ottenere da solo quel tipo di effetto.


Spero che leggerai questo messaggio e che magari possiamo sentirci meglio qui in chat.
Grazie ancora per il supporto e per il tuo ottimo lavoro!



EN
Hey JuanchoTexHD!


I’ve tried all three models you recommended. To be honest, the results didn’t quite convince me, but I want to make it clear that I’m not criticizing your work at all. On the contrary—I really appreciate what you do, and I’m truly grateful for the models. They’re very well made!


What I’m looking for is a more realistic effect, something similar to an example I’d like to show you here in private.
If you could help me achieve something like that, I’d be incredibly grateful!


If it takes extra time or effort to create that effect, please let me know if there’s any cost involved—we can definitely talk about it privately.
Just so you know, I’m still learning how to use ComfyUI and I’m not a professional. I’m working with an RTX 3070 Ti, 16 GB of RAM and 8 GB of VRAM, so I’m not even sure if my setup can handle that kind of effect on its own.


I really hope you’ll read this message and maybe we can chat more in private.

 

[JuanchoTexHD]

View attachment 517227


Ciao JuanchoTexHD!

IT
Ho provato tutti e tre i modelli che mi hai consigliato. Ti dico sinceramente che il risultato non mi ha convinto del tutto, ma te lo dico con il massimo rispetto: non è assolutamente una critica negativa. Anzi, apprezzo davvero tanto quello che fai e ti ringrazio di cuore, perché i modelli sono molto belli!


Quello che sto cercando però è un effetto più realistico, nello stile di un'immagine che ti vorrei mostrare qui in privato.
Se riesci ad aiutarmi a ottenere qualcosa di simile, te ne sarei davvero grato!


Se per farlo serve del tempo o un lavoro extra, fammi sapere anche se c’è un compenso: ne possiamo parlare tranquillamente in privato.
Ti dico già che non sono un esperto di ComfyUI, sto ancora imparando. Ho una RTX 3070 Ti, 16 GB di RAM e 8 GB di VRAM, quindi non so se riesco a ottenere da solo quel tipo di effetto.


Spero che leggerai questo messaggio e che magari possiamo sentirci meglio qui in chat.
Grazie ancora per il supporto e per il tuo ottimo lavoro!



EN
Hey JuanchoTexHD!


I’ve tried all three models you recommended. To be honest, the results didn’t quite convince me, but I want to make it clear that I’m not criticizing your work at all. On the contrary—I really appreciate what you do, and I’m truly grateful for the models. They’re very well made!


What I’m looking for is a more realistic effect, something similar to an example I’d like to show you here in private.
If you could help me achieve something like that, I’d be incredibly grateful!


If it takes extra time or effort to create that effect, please let me know if there’s any cost involved—we can definitely talk about it privately.
Just so you know, I’m still learning how to use ComfyUI and I’m not a professional. I’m working with an RTX 3070 Ti, 16 GB of RAM and 8 GB of VRAM, so I’m not even sure if my setup can handle that kind of effect on its own.


I really hope you’ll read this message and maybe we can chat more in private.

Right now im using horizon XL and chroma depending the situation, chroma is the evolution of flux but is heavy resources, horizon xl is good but you need to play with the denoiser depend the size of the ORIGINAL texture to no create aberrations, the sd upscaler its also important, the realesrganx4 plus its the most "global" solution, but the uniscale v2 its more sharper and preserver, and the ultrafart v3 its in specific situation of edges or heavy noise or need of softness, i deprecated the hatgan in place of the 3 mentioned above. keep on mind this are on constant evolution, new models know better how to RECREATE the image, im not the indicated person to ask to REPLICATE, i am searching constantly for TRANSFORMATION tools, not upscalers, bcoz im not limited to originality cage