Hay Joe a few ?s if you hav the time. I am a portrait photographer and I would like to start offering some of my customers some extra AI created images for fun. So I am working on trying to learn all this from the start. Some videos say to install all this takes a lot of HD space. What do you think I would need for HD space to get everything installed I would need to get up and running with this? Also what would you suggest for Models I think I would need?. Would like to take a single image and just have it enhance that one image into a few different versions first a mild one that just makes the face and person look better but in a natural way. Then maybe a cartoon version and sketch and so on. I see something called photomaker would that be a good to start with? Also I see something called Pinokio that says it then lets you auto install all this and more would that be a good item to use for installing all this? And I would like to be able to do this for two people faces at a time for some?. Sorry for so many ?s but i watch many videos and most jump in with everything all ready installed and working so not 100% on how to even get it all installed Thanks.
Hi, I will try to answer those questions that I can. 1. Resources. ComfyUI is hungry for all resources, not just disk space, but runs better if you have a powerful nVidia graphics card, lots of memory and disk space. My Windows 11 computer has 30 Gb RAM, 16 Gb AMD GPU (Graphics) and my ComfyUI folder is currently over 61 Gb and growing. With all that, ComfyUI can still make my PC crash from time to time, usually because of memory. My PC has AMD graphics which are not popular with ComfyUI - those using nVidia graphics cards get much better/faster results. 2. ComfyUI. ComfyUI is an open source programme, written by developers, it is free to use but does not come with user manuals or technical support, so your best friend is RU-vid videos and (occasionally) the developers technical notes on Github.com. However most of these notes are sometimes meant for other developers and do not always include non technical descriptions for us end users. To get going, watch a few RU-vids on 'how to install ComfyUI' and then give it a go. Then do the same for 'how to install ComfyUI Manager' (which is much easier). Once you have those two things installed and running, you can start learning how to generate images from text prompts (TXT2IMG) and later image to image (IMG2IMG). 3. Models. To start with use the XL models available from Civitai.com such as SD XL, Juggernaut XL and Dreamshaper XL. These models work well with images 1024 x 1024 or 1024 x 768 and is a good start point. 4. Portraits. As you are interested in adjusting portraits, then you will need to master IMG2IMG, but manipulating an image so it resembles the original image with a few changes made is not beginner stuff in my opinion. I have been using ComfyUI for a few months now and still can't do that (yet). I am also a keen portrait photographer so am looking to do similar myself within ComfyUI once I have learned how. Turning an image into a cartoon version is easier with the use of LoRa models. Loras are additional smaller models that do a specific task and there are plenty of Loras on Civitai that can do this - search for RU-vids on ComfyUI Lora demos. 5. Other software. I only use Photoshop and Luminar Neo for photo editing and ComfyUI for AI. I do not know the other products you have mentioned. The key to ComfyUI for beginners is to keep watching ComfyUI RU-vids, absorb what you can and don't be disheartened by what you don't understand, it may sink in later or may not even be relevant. Start with basic TXT2IMG, then IMG2IMG then start adding LoRas or new models and just experiment away. When you get around to IMG2IMG, the image you load (if it is not an image that was already generated in ComfyUI) must be a size ComfyUI understands you cannot load just any size image. Resize your image to 1024x768 or 1024x1024 (if using XL models) and save it as a png file and you should be ok for starters. Hope this helps.
