[Fooocus 2.0.50] Variation/Upscale (Midjourney Toolbar) #390
Replies: 48 comments 61 replies
-
|
I just tried Upscale (Fast 2x) and it did not save in outputs folder. I found the image at AppData/Local/Temp/gradio/114e6f62458641f15248de211b2cda96f59d5169/image.png |
Beta Was this translation helpful? Give feedback.
-
|
fixed in 2.0.51 |
Beta Was this translation helpful? Give feedback.
-
|
Thanks it save now |
Beta Was this translation helpful? Give feedback.
-
|
Sorry, can someone tell me how how can I generate my own photo gallery with Foocus? |
Beta Was this translation helpful? Give feedback.
-
|
i cant find my own |
Beta Was this translation helpful? Give feedback.
-
|
can i use my face |
Beta Was this translation helpful? Give feedback.
-
|
Hehehhe🔫 |
Beta Was this translation helpful? Give feedback.
-
|
How do i get consistent images? |
Beta Was this translation helpful? Give feedback.
-
|
wow ! |
Beta Was this translation helpful? Give feedback.
-
|
@lllyasviel |
Beta Was this translation helpful? Give feedback.
-
|
The Great Demon King lllyasviel attempted to push his own FOC Princess down the endless abyss of MJ~!!! |
Beta Was this translation helpful? Give feedback.
-
|
pls help me ! |
Beta Was this translation helpful? Give feedback.
-
|
I made it ! |
Beta Was this translation helpful? Give feedback.
-
|
@darkymtp It could probably also work with passing I have access to 1 PC with AMD GPU (6600 XT), but it's Windows - and PyTorch on Windows still isn't supported by AMD (ROCm issue #2065), so it's not much use, sadly. |
Beta Was this translation helpful? Give feedback.
-
|
It works. |
Beta Was this translation helpful? Give feedback.
-
|
@darkymtp You can also try Fooocus-MRE variant - it uses newer Comfy version, and also allows passing arguments to Comfy backend as launch parameter (without having to do patches like this), simple |
Beta Was this translation helpful? Give feedback.
-
|
Successfully tested, thanks ! |
Beta Was this translation helpful? Give feedback.
-
|
i can't find the Upscale and inpainting in the colab version. where is it? |
Beta Was this translation helpful? Give feedback.
-
|
please follow official link to use colab |
Beta Was this translation helpful? Give feedback.
-
|
This is such a great update! I love being able to upscale right in Fooocus-MRE! And "Very" is quite cool! Thank you so much - your work is amazing! |
Beta Was this translation helpful? Give feedback.
-
|
Unable to use upscale and inpainting options Should I install anything more supporting dependencies |
Beta Was this translation helpful? Give feedback.
-
|
Dear lllyasviel (I have installed in a separate ssd disc, which is not the main 'C' drive - please advice if I should install any additional dependencies, I am already successfully using Automatic1111 in the main "C" disc) |
Beta Was this translation helpful? Give feedback.
-
|
@tmprabubiz Any log, or screenshot of the Fooocus console window? Maybe there are some errors during startup, or when trying to generate image? |
Beta Was this translation helpful? Give feedback.
-
|
This happened when I tried to outpaint right and left
Already up-to-date To create a public link, set |
Beta Was this translation helpful? Give feedback.
-
|
@tmprabubiz I just tried about the same scenario, worked fine for me: Could you try downloading last MRE release (full version, with env - MRE v2.0.19 should be fine), launch with update, and check if it will be still happening? You can also check official v2.0.50 release, if it won't help. |
Beta Was this translation helpful? Give feedback.
-
|
Dear MoonRide303 Thank You for the response. I installed both twice. But the problem is still alive in both v2.0.19 and official v2.0.50. I have not installed any other models for inpaint or upscale, since I don't know about it. Please guide, if I should install anything in addition. Also, please note that I have not created "path.json" and "user_path_config.txt" in both the versions In MRE v2.0.19, I saw the error lines ///missing {'cond_stage_model.clip_g.transformer.text_model.embeddings.position_ids', 'cond_stage_model.clip_l.logit_scale', 'cond_stage_model.clip_l.text_projection'}//// Already up-to-date To create a public link, set In official v2.0.50, I saw the error lines ////missing {'cond_stage_model.clip_g.transformer.text_model.embeddings.position_ids'}//// C:\Users\Winnertmp\Fooocus_win64_2-0-50>.\python_embeded\python.exe -s Fooocus\entry_with_update.py To create a public link, set Note: Now I have installed both the versions in "C" drive in users folder, where I am successfully using Automatic 1111 without any problem. I think, I am doing some basic mistake, so that in "outpaint" once I hit the "Generate" button, the "warning "Connection Errored out appears" and 1006 appears in command box. Kindly guide me to come out of the problem. |
Beta Was this translation helpful? Give feedback.
-
|
@tmprabubiz This "missing" thing can be ignored, it's just an internal debug (see ComfyUI issue #1100). In those logs you've pasted I don't see any real error, it looks like Fooocus (both vanilla and MRE) started properly. Can you check output on the console in the moment you got that Error in the UI? |
Beta Was this translation helpful? Give feedback.
-
|
My version of error can be reproduced, I just tested twice Load and image to Inpaint or Outpaint (beta) Using Fooocus version: 2.0.78.1 MRE |
Beta Was this translation helpful? Give feedback.
-
|
@barepixels I was able to reproduce it, it was MRE specific problem (related to loading prompts) - fixed, you can restart to get updated version. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you for the speedy fix. Am facing another problem with Outpainting! Seem the newer area created from outpainting is blurry, muddy. Am testing different Sampling Sharpness (default is 2) to see if it makes any difference. Best test try on images that don't have bokeh such as a painting confirm loading prompts fixed !!! |
Beta Was this translation helpful? Give feedback.
-
|
@barepixels Yeah, the out painting in vanilla Fooocus appears to be the same way as Fooocus-MRE, they both seam to be pretty blurry. I think that's why MoonRide303 wanted to wait on implementing it into Fooocus-MRE. The Upscaling however, is awesome! While the "Upscale (2x)" is a bit on the slow side, it definitely improves the image as it upscales it, which is quite nice! |
Beta Was this translation helpful? Give feedback.
-
|
@barepixels You can try doing just single area at a time, it might work better this way. And in general inpainting / outpainting are based on patch inpainting model and sampler provided for vanilla by lllyasviel - improving their output quality might require in-depth knowledge of those, so it would be the best if we wait for original author availability here. |
Beta Was this translation helpful? Give feedback.
-
|
Yep best wait for lllyasviel For demonstration, you can see the blurry of the flowers to the left of her hair |
Beta Was this translation helpful? Give feedback.
-
|
@lllyasviel @MoonRide303 This video explains why Foooocus OutPaint's areas are blurry |
Beta Was this translation helpful? Give feedback.
-
|
THe output in the console is just "1006", meanwhile the warning message
slowly disappears. My major doubt is, if I have to install some other
thing to get the support of upscale and outpatient like installing modells
and upscale models.
With Regards
Thangamani Prabu
Ph: 960006676
Mail: ***@***.***
…On Sun, Sep 24, 2023 at 4:55 AM MoonRide303 ***@***.***> wrote:
@tmprabubiz <https://github.com/tmprabubiz> This "missing" thing can be
ignored, it's just an internal debug (see ComfyUI issue #1100
<comfyanonymous/ComfyUI#1100>).
In those logs you've pasted I don't see any real error, it looks like
Fooocus (both vanilla and MRE) started properly. Can you check output on
the console in the moment you got that Error in the UI?
—
Reply to this email directly, view it on GitHub
<#390 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/BARPIH4FOLNACDKS27EQEL3X35VYDANCNFSM6AAAAAA42WF4LU>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
Hello, I have generated a face, I would like focus to create a character with this face in any position, is it possible? |
Beta Was this translation helpful? Give feedback.
-
|
How can you implement upscale using API? |
Beta Was this translation helpful? Give feedback.
-
|
Please find the websocket API definition in the footer of Fooocus. |
Beta Was this translation helpful? Give feedback.
-
|
I found this, but I still haven’t figured it out, there are no exact instructions on how to do this, for generation there are also pressing different interface buttons. |
Beta Was this translation helpful? Give feedback.
-
|
And another question: is it possible to feed a height map and get a generated image with similar contours? |
Beta Was this translation helpful? Give feedback.
-
|
No, this is currently not supported, only binary mask in inpainting (for now). |
Beta Was this translation helpful? Give feedback.
-
|
And how can I generate and upscale a picture in the same run? Couldn't find any way to do so. |
Beta Was this translation helpful? Give feedback.
-
|
This currently is not possible, it's a 2 step process |
Beta Was this translation helpful? Give feedback.
-
|
Any plans on implementing this kind of feature? I would love to upscale right after generations |
Beta Was this translation helpful? Give feedback.
-
|
Hello everyone ! First of all, thank you for this magnificent software, it's just incredible and stunning at the same time. I have a question : Thanks and sorry for my bad English |
Beta Was this translation helpful? Give feedback.
-
|
Currently there are no plans to integrate video generation. If Midjourney integrates a video creation functionality this can be reevaluated. You can animate images with AnimateDiff using ComfyUI. |
Beta Was this translation helpful? Give feedback.
-
|
@mashb1t |
Beta Was this translation helpful? Give feedback.
-
|
Please elaborate and check the open feature requests (in issues). Are you referring to img2img, upscale or IP adapters? |
Beta Was this translation helpful? Give feedback.
-
|
Thank you for the reply. If I may ask, could you add a batch processing feature for Fooocus's upscale feature? In SD, specifically in the Img2Img feature, it allows users to designate input and output folders for its upscaler |
Beta Was this translation helpful? Give feedback.
-
|
I know I would love to see a batch upscale where one could sic the upscaler on a folder of images an let it run unattended. |
Beta Was this translation helpful? Give feedback.
-
|
Agree. |
Beta Was this translation helpful? Give feedback.
-
|
No I don’t ordered
…On Tue 23. Apr 2024 at 18:45, Manuel Schmid ***@***.***> wrote:
@peladovictor <https://github.com/peladovictor> you can work with masks
and e.g. only change the background if you don't want to touch the
character but change the situation. This is not possible using only the
seed.
—
Reply to this email directly, view it on GitHub
<#390 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AV7VLOTQ66AV425JK3AEB23Y62F3JAVCNFSM6AAAAAA42WF4LWVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TEMBTGQYDI>
.
You are receiving this because you commented.Message ID:
***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
How to fix the error in upscale 2x? |
Beta Was this translation helpful? Give feedback.
-
|
@e2promo make sure you're on the latest version of Fooocus and file an issue if necessary to go into further details. |
Beta Was this translation helpful? Give feedback.
-
|
Do I need to write prompts when using Vary or Upscale? |
Beta Was this translation helpful? Give feedback.
-
|
@dezvin it will improve guidance for both, so it's strongly recommended |
Beta Was this translation helpful? Give feedback.
-
|
Thank you very much for answer! By the way - your fork is top!🔥 |
Beta Was this translation helpful? Give feedback.
-
|
Translation by DeepL. |
Beta Was this translation helpful? Give feedback.
-
|
there is an option "override denoising strength", It worked for me many times. You can check out in this video. |
Beta Was this translation helpful? Give feedback.
-
|
@Kanape21 you may also add the direction the person is looking to to the prompt and then upscale, so thr model knows which perspective to generate. |
Beta Was this translation helpful? Give feedback.
-
|
hey!! can anyone explain me how can we do faceswap using this? |
Beta Was this translation helpful? Give feedback.
-
|
This guy's whole series is good. https://www.youtube.com/watch?v=2hW0M-Y1R6k |
Beta Was this translation helpful? Give feedback.
-
|
Hi is there a way to choose the upscaler to use? I see there is a place to drop upscale models in the file tree, but I don't see a way to specify which upscaler to use in the interface |
Beta Was this translation helpful? Give feedback.
-
|
To install the application I don't have those Nvidia drivers and my laptop
won't support it, which I'm using in colab,runtime is easily getting
disconnected before the output is generated
…On Wed, 4 Sept, 2024, 8:14 am Ni-Deshmane, ***@***.***> wrote:
Hi is there a way to choose the upscaler to use? I see there is a place to
drop upscale models in the file tree, but I don't see a way to specify
which upscaler to use in the interface
—
Reply to this email directly, view it on GitHub
<#390 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/BK6TA4RFZZ4K7O7QNTA4RU3ZUZXZJAVCNFSM6AAAAAA42WF4LWVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTANJTHE3TCMY>
.
You are receiving this because you commented.Message ID:
***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
A career-related photo of a young man can be visualized in various professional settings. Could you specify any particular career or appearance for this young man, like the type of job he's in (e.g., corporate, IT, engineering, medical), or details like clothing (e.g., suit, uniform), background, or expression? This will help me create the perfect image. |
Beta Was this translation helpful? Give feedback.
-
|
How to change default upscaler to another? |
Beta Was this translation helpful? Give feedback.
-
|
|
Beta Was this translation helpful? Give feedback.
and others
-
Foocus 2.0.50 implemented image-based toolbars, making it a relatively complete software for generating high-quality images in a seamless experience.
The toolbar is motivated by the Midjourney toolbar:
And in Fooocus, it looks like this
You may have already immediately realized how to use it – you can drag any images in the generated image sets:
(seed of this example is 808079)
Example: the fierce battle between the brave and evil dragon
The prompt is just "the fierce battle between the brave and evil dragon". Fooocus gives beautiful images no matter how long or short the prompt is, just like Midjourney.
This is Vary (Subtle):
(seed of this example is 12345)
This is Vary (Strong):
(seed of this example is 12345)
Upscale
This is an image I get after some
Vary (Subtle)andVary (Strong)operations. Now I will upscale it.This is Upscale (1.5x):
(seed of this example is 12345)
This is Upscale (2x):
(seed of this example is 12345)
This is the image after Upscale (2x):
(click image to see more details)
Upscale (Fast 2x)
The
Upscale (Fast 2x)is very fast and suitable for upscaling 2k or 1.5k images into 4k images.This can be used as a final step. Below image is 4608x3584 using this as a final upscale
Zoom up:
Another random image I get:
Zoom up:
Note that those examples are not cherry-picked. Users can expect better results if they try more random seeds / prompt engineering.
Beta Was this translation helpful? Give feedback.
All reactions