PlayAI drawing, now I finally don't have to worry about beingold yellowThe knife technique is "backstabbed"! Just need a piece from 4 years agoGTX 1650(Video memory 4GB), the AI drawing effect is comparable to the current best open source model SDXL. This was recentlyGitHubA new project that has been ranked TOP 1 for several consecutive daysFooocus, the three-day mark breaks 4k,ControlNetauthorproduced.
GitHub project address:https://github.com/lllyasviel/Fooocus
Before it comes out, everyone wants to run smoothlyStable DiffusionThe latest XL model cannot compete with the 4060ti with 16G of video memory (the native 3060 with 12G of video memory runs very reluctantly).
Unlike other open source AI tools, Fooocus "focuses on the generation itself", not only has low hardware requirements, but also is easy to get started.very friendly——
The whole process does not need to adjust any parameters, click the mouse,Generate an image in 3 steps.
△Picture source Twitter @Photogenic Weekend
Some netizens called out, "This is simply the culmination of Stable Diffusion and Midjourney":
Say goodbye to manual parameter adjustment! Offline, open source and free, just prompt words and pictures to let the magic happen!
Some netizens lamented: Even a novice can exert the effect of the Stable Diffusion XL model very well.
So, what is the actual generation effect of this new image AI tool? We tried it out.
Colab produces pictures in half a minute, and the effect is comparable to SD
From the running interface, Fooocus has a total ofMore than 100 built-in stylesAvailable to choose from.
△Picture source Twitter @camenduru
As forefficiency, Fooocus also produces pictures very quickly. In Colab, it takes about half a minute to draw a picture in speed priority mode:
Then let it draw firstcomicsTry it to see what the "Mazar War" will look like from the perspective of AI. (Not this Mazhaha)
Since AI is used to directly generate portraits, there are still some problems with hand processing, so we simply asked Musk and Zuckerberg to wear gloves:
The effect seems to be pretty good. I don't know if they have any bets, but they might as well let the loser dress up as a woman.
(Reminder: There is no winner in a fight)
In the end, the two "shaked hands and made peace", and this precious scene was also recorded by the photographer. Does the overall scene look like that?
After the "Battle of Mazar" ended, Lao Ma obediently returned to the company and sold Tesla.
neglectLOGOif,posterThe design sense is also quite online.
In fact, every built-in style of Fooocus is very interesting, so let's take a look at thesedifferent stylesThe live picture of:
In terms of imitating famous works, there arecyberpunkVersion,ZeldaVersion,Minecraftversion, or evenPokémonversion of Musk is available for viewing.
As for other art forms, there arePixelandLowpolystyle, andNendoroidandpaper cutVersion……
Of course, there are endless examples, readers and friends of more styles can experience it by themselves.
(I have to say that the text in the AI drawing is finally not a ghost drawing)
Are the great painters already impatient and want to try it? We'll introduce how to play Fooocus right away!
The interface of Fooocus belongs to Aunt Jiang, which can be said to be very concise:
If you just have a taste and have no special requirements,This prompt box is completely enough.
Because the author has internalized many complex skills into the program, these operations of tuning parameters no longer need to be done manually.
Enter the prompt directly in the box below, click the generate button and wait for the drawing to be produced.
(By default, two pictures are output at a time, the size is 1152×896, the style is cinematic default, and the speed is prioritized)
If you need advanced settings, check Advanced in the lower left corner, and the configuration information will appear on the right side of the page, divided into three tabs:
What can be adjusted includesSize, quantity, style, performanceetc.
If you are a professional player, you can also choose a model version and even adjust LoRA parameters.
Beside this there isAdjust sharpnessThis kind of high-end gameplay.
For the same content, the following animation shows the change in sharpness from 2 to 10 to 20. It can be seen that as the sharpness increases, the details of the picture become richer and richer:
However, whether Fooocus supportsChinese, we also tried it, but unfortunately it doesn't work yet.
For example, we input the prompt word "apple", and the result is a girl.
Does this…do you want to say "You're the apple of my eyes"?
Now you probably know how to play Fooocus, so how to configure it?
If you have an NVIDIA graphics cardwindowsdevice, you can use the out-of-the-box version. (This is probably the 114,514th time Lao Huang has won pot)
At the same time, the hardware needs to meet the minimum configuration requirements-4GB video memory + 8GB memory.
Download directly from here first:
https://github.com/lllyasviel/Fooocus/releases/download/release/Fooocus_win64_1-1-10.7z
After decompressing, double-click run.bat to run. The system will automatically download the model and deploy it. It can be used after the configuration is completed.
linuxThe configuration requirements of the version are the same as Windows, but the configuration process is more complicated.
(If you have Jupyter, you can also refer to the note files used in Colab)
First, install the environment dependencies:
<span class="hljs-attr">git</span> <span class="hljs-string">clone https://github.com/lllyasviel/Fooocus.git</span> <span class="hljs-attr">cd</span> <span class="hljs-string">Fooocus</span> <span class="hljs-attr">conda</span> <span class="hljs-string">env create -f environment.yaml</span> <span class="hljs-attr">conda</span> <span class="hljs-string">activate fooocus</span> <span class="hljs-attr">pip</span> <span class="hljs-string">install -r requirements_versions.txt</span>
Then download the model file and store it in the specified directory:
△Please see the GitHub page for details
Of course, you can also let the system download the model automatically:
<span class="hljs-selector-tag">python</span> <span class="hljs-selector-tag">launch</span><span class="hljs-selector-class">.py</span>
if you are usingmac, or the hardware configuration does not meet the requirements, it can also be run directly with Colab.
(Portal: https://colab.research.google.com/github/lllyasviel/Fooocus/blob/main/colab.ipynb)
However, I still have to complain that the Colab version will crash several times from time to time, either automatically stopping or memory overflow…
If you want to run Fooocus more smoothly on Mac or A card computer, you can wait for the author's update.
Overall, Fooocus produces good pictures. If the prompt words are chosen well, it can even be used as Stable Diffusion. The key is that the hardware configuration requirements are not high.
How is this achieved?
Latest projects from ControlNet authors
In terms of architecture design, Fooocus is mainly divided into two parts: the interactive interface and the AI model.
in,user-interfaceTwo projects are referenced, namely stable-diffusion-Web UIand ComfyUI.
stable-diffusion-webui is mainly the front-end design of the interactive interface:
ComfyUI combines Stable Diffusion's GUI and back-end design:
As forAI model, it can be seen that the new SDXL model of Stable Diffusion is used:
This is currently one of the best versions of Stable Diffusion, and the generation effect has been improved a lot compared to the previous version 1.5.
However, although the model and UI design of Fooocus refer to the ready-made Stable Diffusion open source project, the author incorporated a lot of his own optimized designs during production to make the model run smoother.
For example, the author carefully adopted the self-designed k-diffusion sampling method in Fooocus, which can improve sampling continuity, reduce performance loss, and improve sampling efficiency;
In addition, the author also carefully adjusted the parameters of the sampler (Sampler), and modified and added some new settings including movie style based on the original version.
The reason why the LoRA option is added to Fooocus is because the author's tests found that the SDXL model with LoRA (weight setting less than 0.5) is almost always better than the SDXL model without LoRA.
The author who developed the project Fooocus is namedLvmin Zhang, graduated from Soochow University in 2021, and is currently a PhD student at Stanford University.
Several projects he has worked on, including ControlNet and style2paints, almost all have become popular:
Now, the latest project, Fooocus, seems to be equally popular.
On social media, some netizens have spontaneously compiled a collection of different styles of prompt words for Fooocus.Excel.
If you don't know what kind of picture to generate, just refer to the prompt words in this document:
Have you figured out what kind of images you want to generate with Fooocus?
project address:
Reference link:
