🏆🏆🏆 MY BOUNTY IS LIVE 🏆🏆🏆 Looking for BEST images for Cinematic Realism!
https://civarchive.com/bounties/10855
This workflow UNLOCKS the full potential of Z-Image. It generates pixel-level fine realistic details with very fast gen speed, for both foreground and backgrounds!
I spent months researching and experimenting Qwen hyper realism and now weeks of chasing after Z Image realism, and I'm now sharing my full workflow with this release, you might find it useful to use.
Thanks for the shoutout by Z Image dev team TongyiLab!! https://x.com/Ali_TongyiLab/status/2000762326136578121
v2.2 Key features: (see compare pics in v2.2 post, not going to make v2.3 until I find another significant image quality boost technique)
NEW Detail vs Cinematic master controls. Now users have full control over this!
NEW Quick reuse prompt section, one click regen image with new seeds
NEW Figured out Z Image fp32 noise issue, attached fix suggestions
Achieve realism across broad range of fictional/nonfictional prompts
Get high details up close AND far away in background
Maintains realistic skin/texture through main image and through upscale
Very fast initial generation, upscale optional.
Uses my Loras to achieve higher detail density & improved composition
Built in high quality+fast dual workflow/stand-alone upscaler
High quality Upscaled Inpainter
Realistic PBR like having nice specularity and sharpness of edges and good metalness
Easy to use workflow that won't require per-image tweaking
Photographer style Neutral image output with max color range
Clean 3D structure such as building lines and objects
❤And if you want to incorporate some parts into your workflow, I would love a mention too! 🏆MY CINE REALISM BOUNTY IS LIVE: https://civarchive.com/bounties/10855
Note: This workflow works much better with my Cinematic and CyberHD Enhancer loras since they increase the resolution of the details and textures.
FLOWMATCH NOISE FIX: ComfyUI beyond 0.3.77 changed the default UNET torch weight dType, for many users it used to default to fp32. FlowMatch is sensitive to this, and works best with fp32. To set the toggle, edit your .bat file for launching ComfyUI, and add to the end: --fp32-unet
Comparison article here: https://civarchive.com/articles/24424
IF YOU ARE NOT GETTING MY LEVEL OF DETAIL/NOISY IMAGE:
It probably means your Flowmatch isn't working. You MAY get this error log: The config attributes {'stochastic_sampling': False} were passed to FlowMatchEulerDiscreteScheduler, but are not expected and will be ignored. Please verify your scheduler_config.json configuration file.
(I had to downgrade to ComfyUI 0.3.77 to work)
(SECURITY ALERT: DEC 29: FIXED in my Workflow v2.01 REMOVED: RMBG nodes from AILAB. Security Vuln in nodes.
Usage Guide:
Just toss in a prompt and it'll just work!
Let me know if you have any questions/guidance requested in comments.
Description
v1.8 Please read section notes!
Major retune, images will now get a lot more detail and anti aliasing quality.
NEW SMOOTHNESS full range settings.
Improved object structures when applying texture noise.
NEW high speed high quality upscalers, settings config for both low or high vRam (standalone and workflow versions)
Added CLIP model unloading after final Positive Clip Encode of the workflow, to avoid OOM during upscaling.
FAQ
Comments (40)
The details of version 1.8 are good. However, the afterimage happens as if there are two films overlapping. It will be better if you adjust the details, but it's a shame that the reason for using z-image is that it has a speed aspect. For your information, I use a dype node
The ghosting occurred after some final tweaks I did right before release, I'm rolling back to find the bug and fixing it before releasing v1.8b, thanks for feedback!!
v1.9 is LIVE!!! Thanks for being patient!
Fixed ghosting error and retuned image for quality while I was at it. Organized WF a bit horizontally, easier for new users. Lots of upgrades from v1.7 since this is merge of v1.8 and v1.9.
respect, you finally succeeded 👍👍👍 so far I can't find anything that is bad
v1.91 LIVE. This fixes the too-strong-details. Some settings made it very easy for tiny pellets to appear in image, this version fixes it.
This is an exceptional workflow. I really appreciate the comments/notes. Thank you for your contributions to the community.
You are welcome! I will have 2 more major updates and then the workflow should be "done". Version 2.0 and version 2.1. Version 2.0 is another large improvement in visual quality, should allow for fully complex and detailed elements across entire image. From full new york building cityscape, to a vast army fighting in a sea of grass. Version 2.1 is going to be upgrading the Inpainter section with some automation. And then it'll be done for a while until there's any further significant upgrades. v2.0 is the big one! (it's in final beta testing at the moment, I'm adding in a "cinematicness" configuration as its final feature)
v2.0 Final: today's release candidate testing going well, last few days had some set backs that required retuning. Because each release candidate requires testing 100+ prompts, when some fail to meet quality requirements, I go back and retune everything again. I do this so that the broadest range of users/art styles will be able to use this workflow easily without adjustments needed. Will do another overnight batch to do final testing and hope it goes well!
come on
@zq1987 it aborted about 3 images after I left the keyboard, so have to do the 100 prompts again today.
Another big retune today to improve quality even for bad prompts, now running 250 prompts to test it overnight.
260+ prompts went well, came out with 85 that are now good enough to be in my personal "to be posted" folder. That's a very high %, so the v2.0 workflow is ready for release. I'll just need to prep some things and then it'll release today/tomorrow.
Despite all my attempts, I have been unable to integrate i2i into my existing working system to increase detail. Could you help me with this?
Just use my workflow? =)
I have to use an SDXL-based workflow (including brushnet), and I want to use the image I obtain as a result of this workflow for detailing. Your workflow is truly flawless, congratulations.
@cemilhaci2959 ok, so you mean you want to output my workflow's image, and then output it to SDXL refining step for refinement/upscaling? You would just... output the pink Latent, and then plug it into SDXL sampler, with 0.2 to 0.5 denoise... what are you getting stuck with?
On the contrary, I am considering taking my SDXL output and improving the quality with your workflow.
@cemilhaci2959 ahh I see. My workflow is carefully tuned on the first sampler progress (even more so in the upcoming v2.5) but you can combine your SDXL outputs by replacing my sampler1 with your sdxl image, then testing different denoise level on sampler2.
thanks bro
I am not sure if this question has been answered, but I need some guidance on using the luts. I put the downloaded lut inside the comfy ui essentials lut folder, but i did not have any success. Also, i tried creating a lut folder in the models directory with no success. What other options are there?
The LUTs should be placed inside the "custom_nodes/comfyui_essentials/luts" folder
@luneva interesting, i am not having any luck with the lut folder in that location. maybe it is the lut file? i went to the website provided. is there a special task needed to active the downloaded lut file?
@Noob_ee after y ou download them, you have to press R in comfy, or refresh the browser page, to start seeing the newly downloaded LUTs.
@luneva Never mind, it is working now for some reason, most likely my mistake. Thank you very much.
@Noob_ee I rarely use LUTs at full strength, so try strength such as 0.2 or 0.3 or even 0.5.
@luneva i think ill bypass it, i am generating an error. maybe a bug in the code. thanks for the tip. great workflow.
@Noob_ee yes I've been experiencing weird conflicts between nodes lately. I've addressed as much of those as possible in the coming v2.0 workflow though! Will probably drop it today or tomorrow.
@luneva awesome, so not so much a user error, haha. im glad that this workflow can still operate without certain sections. im looking forward to the knew version.
@Noob_ee me too, this v2.0 took forever to tune. Here's a sneak peek at an output: https://imgur.com/a/TpP165Q
@luneva i was very discouraged looking at the wrokflow, but i do have some experience now with comfy ui, took me a while to adjust. i like the GUI because it makes thinking low, lol.
@luneva that is an amazing image, i bet that cafe serves great beverages. im glad i came across this workflow, it is going to do wonders for my project. it is making things much smoother. i am not experienced with inpainting, which will definitely come in handy for those final touches. i do use other platforms like Krita to handle the editing.
@Noob_ee I'll update the inpainter section in 2.1 with optional automation, like if someone wants to face-fix EVERY image. My inpainter section doubles the scale, so you'll always get better details with it, whether face fixing or hands/feet fixing.
@Noob_ee I got yelled at for making a messy WF in the earlier versions, so I cleaned up the workflow and made it linear (from left to right), and you don't really need to adjust any settings the way I've tuned it, it's meant to be one-click-done. ( just add prompt) But my notes are useful to read though for best results!!
@luneva i agree, the workflow is very clean and easy to follow. even better if it is already for one click and fire. the only part ive messed with was the seed to random to change the image. eventually, ill play with control net and inpaint.
@Noob_ee I haven't used the controlnet section at all after I implemented it, so its nothing special. The inpaint section should work pretty well though. have fun =)
@luneva will do thanks.
I'm eagerly waiting for 2.0.
Thank you for sharing this workflow.
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.