Hello, this is PD Jo. Today I will explain to you the easiest way to create AI images. When discussing AI image generation, everyone talks about stable diffusion and midjourney. The advantage of table diffusion is that it is free. The advantage is that you can create images privately. The advantage of Midjourney is its simple operation method and quality results. The AI tool we will introduce today has both of the above two points. It doesn't require much computer specs either. Today’s AI tool is Focus. How should I pronounce this? It's focus, but it's also focus. Then let's begin First, access GitHub through the link in the description section. If you scroll down a bit, there is a download link. Click to download If you unzip it, there is a run batch file Click the batch file to run it When you run it for the first time Because models are downloaded automatically It takes a little more time This is the executed screen how is it The UI is overly simple. The prompt and the generate button are done. This is where you can really feel the philosophy of focus starting from the UI. Let’s simply enter the prompt and test it out. What do you think? It's really awesome. I entered only one word at the prompt. An almost photorealistic image was created. This time, we will create an image of a person. I think it came out so beautifully. Now let’s look at other options Check the input image below Select the option on the right You can vary or upscale Disable does not refer to the image. Create again by referring only to the prompt. You can think of Very Subs and Strong as being distinguished by the degree of image change. The three upscales are differentiated by size and speed Click on Inpaint or Outpaint below. This is the same function as Stable Diffusion’s Inpaint. Drag the image created above into the expanded area. Apply a mask to the ears Enter the earrings in the prompt Click the Generate button Earrings were added to the existing image. The earrings came out nicely too. Now let’s look at more advanced options. Check Advanced here As the UI on the right expands Options are available First of all, you can choose speed and quality in performance. The difference between the two is Speed is generated quickly because the sampling step is 30. For quality, the sampling step is set to 60 to provide more detailed expression. Below that you can select the resolution and ratio of the image. The image number below is the quantity of images created simultaneously. Below that is the negative prompt input field. And if you uncheck the random check at the bottom, an input field will appear where you can enter the seed. The Settings tab consists of only the most used functions. Let's look at the styles of the next tab. As you can see, there are so many styles. Styles can be selected multiple times I will test it and recommend some good styles. comic book style fantasy art line art neon punk retro cyberpunk dark fantasy pop art two There are so many that I haven't tried them all yet. Below are the model options You can choose between a base model and a refiner. Below you can choose rollers and weights Base models and refiners can be added to the checkpoint folder in Models under the focus folder. You can add the roller file to the roller folder in Models under the focus folder. If you add model and roller files, click Refresh All File once and then You can choose Model and roller files are only available in SD XL version. Next is Advanced Sampling Sharpness: The higher the value, the sharper the images and textures will be. If the sharpness value is different, the image will look like this: The higher the value on the guidance scale, the cleaner, more vivid and artistic the style becomes. If you adjust the guidance scale value, The image changes like this Lastly, check Developer Debug Mode to further expand your options. But at this point, this thought occurred to me. The key to focusing is to create simple, high-quality images. Do my subscribers want such complicated options? I will boldly pass this part. Next, I changed the model and created a few images. The model used dream shaper I tried Focus today The biggest advantage is that it can create simple yet high-quality images. And the most important thing is that it runs well even on low specs. I hope many people will try this great tool. That's all for today's video If you found it useful and fun, please like and subscribe. We will come back with more advanced videos in the future. thank you
