JonXL's Minimax H3 Image Upscaler (LORA + Workflow & Batch)
Consider to download the workflow (from 'optional downloads') for easier image extraction and batch processing.
A conceptual lora and workflow to upscale blurry and pixelated images up to 1920x1920 with Minimax H3.
Extensive testing and multi pass training made it now possible to upscale images with Minimax-H3.
The workflow can take a single image, or take a batch input folder and will automatically upscale image(s) to the nearest given resolution up to 1920x1920.
The upscaler seems to work very well for blurry images taken from a video clip or heavily compressed photos.
(!) Note #0: There are now two LoRas, the 24 Frame Quality lora (1920) and the 1 Frame Speed lora (1536 1 frame) . Each has there speed / quality trade offs. Make sure to download the workflows from the relevant page. (1 Frame workflow requires a custom node, see bottom of the page)
(!) Note #1: This is an upscaler, not a detailer. It will not add more details, but it can extract a lot of detail from blurry and unsharp images. Don't expect it to add detail to already sharp / AI generated content.
(!) Note #2: For maximum quality use the 1920x1920 resolution and the 24 Frame Quality Lora for best results. This can take up to 5 minutes on a RTX5090. The 1 Frame Speed lora however can generate within 30 seconds on a RTX 5090 at the cost of (detailed) quality. Consider lowering your resolution to 1280x1280 for the quality lora, if you have a lower hardware specs or less patience.
(!) Note #3: You can enable the batch mode option in the AIO workflows, supply an input and output folder and it will automatically upscale all images for you, even keeping the file names.
Prompt:
A static still blurry pixelated picture crossfades into a clear sharp detailed picture, 4psc4l3,
// always use this prompt!
Authors comment
Feel free to post your safe for work results on this LoRa's page too!
Please enjoy upscaling your content with this upscaler lora and workflow!
Usage:
Use the following trigger sentence to activate the image generation
A static still blurry pixelated picture crossfades into a clear sharp detailed picture, 4psc4l3,
Seems to work very well on resolutions of 1280x1280 and 1920x1920
It will auto-scale the input image to the nearest resolution
Default values of 24 frames for Quality lora, CFG 1 and Lora strength one should be kept
You can change the CFG and frame count, it might add more detail but it's not recommended unless you like to experiment
For the 1 Frame Speed Workflow, you can use the Turbo Lora to improve contrast at the cost of some consistency, or disable the Turbo Lora for more consistency. I recommend keeping the Turbo lora enabled.
When using the 1 Frame workflows, make sure to clone the custom node and download the special VAE, no turbo lora required.
Technically, you can also use this in Text2Video mode, and generate photorealistic imagery. It was trained on the same dataset as https://civarchive.com/models/2884206/minimax-h3-photorealistic-image-generator-lora-workflow
Let me know your feedback
Known issues:
You may need to install some frequently used nodes, mostly Kijai's nodes and VideoHelperSuite
Description
FAQ
Comments (21)
The workflow is a bit of a noodle nightmare lol but srsly bro this is really nice, I turned up the frames from 24 to 48 and the results are really nice for recovery on some real old bad quality images. Tks for sharing this
Thanks! Happy to hear your feedback!
You will probably like the batch feature too :)
I try most ai upscales and this is already one of the best! It does a great job of staying true to the original details. I might suggest a much simpler single image workflow to attach. Great work and thanks!
Thank you for your feedback and liking the model! I'll take the suggestion of making a simpler workflow into consideration :)
I'm trying this upscaler after reading your comment.
Can you name your top upscalers besides this one, please?
Thanks in advance.
@nightcall You mean outside H3 (as this is the only image upscaler for H3) ?
SUPIR for images is pretty decent and FlashVSR/ SeedVR2 for video is also decent.
@JonXL Thanks for the suggestions. I just wanted to try more upscalers and compare them with this LoRa.
This one is excellent for details, but also changes brightness, color tone, and causes pixel shifts that slightly change the aspect ratio.
@nightcall Have you tried both loras, both the 24 Frame and 1 Frame lora? The 24 Frame is a lot better than the 1 Frame
Its quite good at upscaling blurry photos. I like the results more than SeedVR2 tho its overall also slower due to having to generate multiple frames.
I got the workflow working but as others have said it would be nice if you could make a version with less custom nodes/simpler.
Hey Tosermepis,
Thanks for your feedback and appreciation!
I'll consider making a mimimalist workflow.
Regarding the multiple frame generations, it seems H3 produces stable image output starting by default around 22 frames. It therefore needs the overhead unfortunately.
I am experimenting the coming days making a 5 frame upscaler lora that uses the MM 1 image node. That 5 frame lora will probably be a "turbo" version, 4 times aa fast but at the cost of some quality.
If you want you can now try out the new 1 Frame Speed lora (has its own workflow).
It's almost 10 times faster on a 5090 (less than 30 seconds) but the upscale quality is more basic.
@JonXL Thanks, will definitely try it out for my next upscaling
I'm shocked at how well this works to unblur images. I have some slightly blurry images that SeedVR2 can make crisp and sharp, but they wash out detail like freckles. This works astonishingly well for stuff like that.
Thank you for your feedback jjjshabadoo!
Yes, I am quite amazed by the results myself too ;)
As of 10.09.2026 I updated the workflows (due to request)to now include an 'Easy' worfklow that is more basic than the workflow posted earlier (now called AIO)
https://civitai.com/api/download/models/3309188?fileId=3200114
Oh! If it can be used to improve video clarity, it would be revolutionary
Hey there!
Yeah, that's a whole other league though :)
I have some ideas but it's exponentially more complex and computational intensive.
just add it through lora loader and you good to go
As of September 13th, there are now two LoRas,:
- The 24 Frame Quality lora (as originally posted)
- The 1 Frame Speed lora.
The 24 Frame Quality Lora has superior upscale quality but can take 3-5 minutes to upscale on a RTX 5090.
The 1 Frame Speed lora is a bit of a ' turbo' lora, it is about 10 times faster on a 5090 but at the cost of fine detail
If you just need to unblur and somewhat sharpen your images consider using the 1 Frame Speed lora, but if you need best quality (preserve skin details) then use the 24 Frame Quality lora.
Each requires there own workflow, download it from the related page.
When using the 1 Frame workflows, make sure to clone the custom node and download the special VAE.