How To Make Photorealistic Images In Fooocus

Monzon Media
10 Dec 202306:03

TLDRIn this tutorial, the presenter unveils a straightforward method for creating photo-realistic images using Focus, a platform that leverages Stable Diffusion 1.5 models. By selecting the appropriate Advanced tab settings, users can experiment with different styles and models, such as Juggernaut XL and Realistic Vision version 6, to achieve a hyper-realistic or photo-realistic look. The video demonstrates the process by comparing the results of using a standard Stable Diffusion 1.5 model versus the described method, highlighting the advantages of higher resolution and refined image quality. The presenter also shares tips on where to obtain the necessary models and how to optimize settings for various image types.

Takeaways

  • šŸŽØ The video demonstrates a method to achieve photo-realistic images using Focus with specific settings and models.
  • šŸ”§ Ensure the 'Advanced' tab is checked to access the necessary options for this method.
  • šŸ–¼ļø Under 'Styles', select 'Focus V2', 'Focus enhance sharp', and 'photograph' for optimal results.
  • šŸŽ„ Occasionally, switching between 'cinematic' and 'photograph' styles can yield different looks.
  • šŸŒŸ Utilize 'Juggernaut XL' as the base model for this particular setup.
  • šŸ” For the refiner, use 'Realistic Vision' version 6, which is a stable diffusion 1.5 model.
  • šŸ“ˆ The refiner setting at 8 is recommended for Excel refiners.
  • šŸ› ļø The method is a 'hack' to combine the benefits of stable diffusion 1.5 models with the Focus platform.
  • šŸ“ø This approach avoids common issues with stable diffusion 1.5 models at higher resolutions, such as deformations.
  • šŸš€ The script mentions 'sdxl offset Laura' with a weight of 0.5 as an additional option to experiment with.
  • šŸ  Models can be obtained from Civit AI, including 'Realistic Vision' version 6 and other photo-realistic models like 'Epic Photogasm' and 'Epic Realism'.
  • šŸš« The video creator advises against using 'Extreme Speed' as it may not work well with the turbo models.

Q & A

  • What is the main topic of the video?

    -The main topic of the video is to demonstrate the easiest way to create photo-realistic images using Focus with specific settings and models.

  • Which tab in Focus should be checked first?

    -The Advanced tab should be checked first to access the necessary options for creating photo-realistic images.

  • What are the available styles under the Advanced tab in Focus?

    -The available styles under the Advanced tab are Focus V2, Focus enhance sharp, and photograph.

  • What model is the presenter using as the base model in the video?

    -The presenter is using Juggernaut XL as the base model for creating photo-realistic images.

  • Which refiner model does the presenter use and why is it significant?

    -The presenter uses Realistic Vision version 6 as the refiner model because it allows for utilizing the benefits of a stable diffusion 1.5 model within the Focus platform.

  • What are the advantages of using the described method over generating a stable diffusion 1.5 model directly?

    -Using the described method allows for higher resolution images without the common issues like deformed limbs or double heads that can occur with direct generation in certain UIs.

  • Where can users find and download the mentioned models?

    -Users can find and download the mentioned models on Civit AI's website.

  • What are some of the photo-realistic models available on Civit AI?

    -Some of the photo-realistic models available on Civit AI include Realistic Vision version 6, epic photo gasm, and epic realism.

  • What are the negative prompts used by the presenter and why are they used?

    -The presenter uses minimal negative prompts such as 'ugly', 'deformed', 'text', and 'watermark' to avoid undesired features in the generated images.

  • How does the presenter test the effectiveness of the method?

    -The presenter tests the effectiveness of the method by generating images with different base and refiner models, comparing the results side by side, and showcasing various examples on screen.

  • What is the presenter's conclusion about the method?

    -The presenter concludes that the method is effective in creating photo-realistic images, works well for various subjects, and can be further optimized by experimenting with different base and refiner models.

Outlines

00:00

šŸŽØ Photorealistic Image Creation in Focus

This paragraph introduces a method for creating photorealistic images using Focus, highlighting the importance of the Advanced tab and specific options under Styles such as Focus V2, Focus enhance sharp, and photograph. It explains the use of Juggernaut XL as the base model and Realistic Vision version 6 as the refiner, emphasizing the stability of the diffusion 1.5 model. The paragraph also discusses the benefits of using this method over standard stable diffusion 1.5 models, particularly for higher resolution images without common artifacts. Additionally, it provides information on where to obtain the models and how to set them up in the software.

05:01

šŸš— Achieving Photorealism in Various Subjects

The second paragraph discusses the versatility of the method in creating photorealistic images of various subjects, including cars and people. It shares the creator's personal experience with using face swap in Focus and how the method helped improve the likeness. The paragraph also touches on the potential for further refinement of settings to achieve a closer representation of the creator's appearance. It concludes with a mention of a previous video on face swapping and encourages viewers to check it out.

Mindmap

Keywords

šŸ’”Photo-realistic images

Photo-realistic images refer to digital creations that closely resemble real-world photographs in terms of detail and visual fidelity. In the context of the video, the presenter aims to demonstrate a method to consistently produce such images using Focus, a platform for generating images from text prompts.

šŸ’”Focus V2

Focus V2 is a version of the Focus platform that the presenter uses to enhance the sharpness and photographic quality of generated images. It is one of the options under the 'Styles' tab in the software, which is essential for achieving the desired photo-realistic look.

šŸ’”Stable Diffusion 1.5

Stable Diffusion 1.5 is a specific model used in the image generation process that is known for its ability to produce high-quality, photo-realistic images. It is used as a refiner in the described workflow, contributing to the final output's realism.

šŸ’”Base model

In the context of the video, a base model refers to the primary model used in the image generation process. It serves as the foundation upon which other enhancements and refinements are applied to achieve the final image.

šŸ’”Refiner

A refiner is a secondary model used in the image generation process to fine-tune and improve the output based on the base model. It helps to achieve specific visual effects or qualities, such as photo-realism.

šŸ’”SDXL pipeline

SDXL pipeline refers to a specific image generation process or workflow that is optimized for using Stable Diffusion models. It is mentioned as a key component in the video's method for achieving high-resolution, photo-realistic images without common artifacts like deformations.

šŸ’”Civit AI

Civit AI is a platform mentioned in the video where users can download various models, including the 'Realistic Vision' and other photo-realistic models like 'Epic Photogasm' and 'Epic Realism'. These models are essential for users looking to replicate the video's workflow.

šŸ’”Dream Shaper XL Turbo

Dream Shaper XL Turbo is a base model mentioned in the video that is used for generating images. It is noted for its potential in the workflow but currently has some limitations, such as not being optimized for 'Extreme Speed'.

šŸ’”Negative prompts

Negative prompts are specific instructions given to the image generation platform to avoid including certain undesirable elements in the output. In the video, minimal negative prompts are used, such as 'ugly', 'deformed', 'text', and 'watermark', to ensure better image quality.

šŸ’”Face swap

Face swap is a technique where a person's face is replaced with another's in a photograph or video. In the context of the video, the presenter humorously references a previous attempt at face swapping using Focus, which did not yield the desired results but was improved with the described method.

Highlights

The presenter is showcasing the easiest way to create photo-realistic images using Focus.

Ensure the Advanced tab is checked to access the Styles options.

Under Styles, you can select Focus V2, Focus enhance sharp, and photograph.

The presenter sometimes swaps between cinematic and photograph for a slightly different look.

For the base model, Juggernaut XL is used, which is crucial for achieving photo-realism.

Realistic Vision version 6 is used as the refiner, a stable diffusion 1.5 model.

Focus is primarily an sdxl platform, but this method allows for the use of stable diffusion 1.5 models.

The refiner setting for animame models is recommended to be at 4 or 6.

Excel refiners can be set to 8 for optimal results.

The sdxl offset Laura with a weight of 0.5 is suggested for the best results.

The presenter demonstrates the differences between using a stable diffusion 1.5 model and the described method.

The method avoids common issues like deformed limbs and double heads at higher resolutions.

The presenter provides examples of where to get the models, such as civit AI.

Realistic Vision version 6 and other photo-realistic models like epic photo gasm and epic realism are mentioned.

The presenter shares their personal experience with the method, including face swap results.

The method's ability to produce high-quality results for both people and objects like cars is highlighted.

Multiple combinations of base and refiner models can be experimented with for varied results.

The presenter concludes by encouraging viewers to check out a previous video on face swapping with Focus.