Create Consistent Characters or Just About Anything In ComfyUI With Gemini 2.0 Flash ( Zero GPU)

okay so today we're going to dive into the easiest way to create consistent characters or honestly just about any type of image in comfy UI and here's the thing we're going to use a node that does exactly what I wanted omn gen to do it's simple it's practical and it's 100% free all you need is an internet connection oh and you don't even need a massive graphics card to make this work how cool is that all right let's get started but before we do if you're into the wild Tangled spaghetti that is comfy give it a like And subscribe for more videos trust me you'll want to stick around for this so here's the workflow we're looking at today it's super simple but don't let that fool you it's incredibly effective and the best part you can use it alongside other workflows the workflow itself has just two main groups an image input group and the star of the show the llm group for this demo I'll be using Google's new Superstar gemin 2 flash for both image and text generation let me walk you through how this works first you upload an image into the image input group then using gemin2 flash you can tweak that image remove something from it or generate something entirely new honestly your imagination is the only limit here the image input group consists of for load and resize nodes I chose these nodes specifically because of the resize function if the image you upload is too large you can resize it by enabling the resize option and adjusting the dimensions just remember to disable resize if your image is already small enough you'll see the size of the image displayed right below the node so it's easy to keep track now I've also included a load batch image node directly below these I haven't tested it yet but I thought it'd be good to have it there just in case here's how the image input works there are four upload nodes and they're controlled by a switch don't worry it's super straightforward the switch goes from 1 to 5 if you're only using one image loaded into the node labeled one and set the switch to one if you're using two images set the switch to two and upload two images this is for the first two images same deal for three or four Images if you're using three images set the switch to three and upload three images to the first three upload nodes and for four Images the switch should be at four just set the switch to the corresponding number and you're good to go if you set the switch to five it automatically disables the load and resize image nodes and enables the batch image upload instead and there's also a bypass image input option which mutes the image input entirely if you want to use text only generation neat right now let's move on to the large language group this is where the magic happens we're using the if llm node here let's quickly set it up before we start the node itself can use either o or a bunch of other AI models but for this video we're sticking with gemin for the model I'm using gemin flash but sometimes you might only see the Quin models listed if that happens here's what you do select another provider then go back and select gemin again this should refresh the list and show all the models supported by gemin from there select gemin 2 Flash 12. 06 and make sure the strategy is set to gem to create next up you'll need to enter the API key don't worry I'll delete this API key as soon as I'm done with the video and here's the best part the API is 100% free you don't pay a single penny it's like our Tech overlords decided to throw us a bone for once I'll show you how to get the free API later but first let's test the workflow before you dose off all right let's get to the reason we're all here character consistency and creating a Laura data set for a character I'll upload an image of a cute low poly game character what I want is a character sheet with all the views front side and rear I'll set the switch to one since I'm only using one image the resize option is off by default because the image isn't too large the image input bypass is set to yes since I'm using an image as input for the prompt or type generate a front side and rear view of the character it only gave me the front and rear views I'll have to try again this time I'll tweak The Prompt two generate an image of the front view rear view and a side view of the character it seems like it created three images even better this makes creating a data set for Aur so much [Music] easier let me try this with another image I'll use this image to create a set of consistent images the pr prompt will be generate images of the character in different profiles and different emotions and this right here this is what it's all about look at that it's creating exactly what I asked for now let's see what else we can do with this thing this time I'll try multiple images let's say I have a black water bottle and logo I'll set the switch to two for image one and two The Prompt will be a black bottle with the logo printed in the center of the bottle that's simply amazing it even changed the orientation of the bottle that was really a nice touch one last image before I try it without an input image I'm going to let gem create a YouTube thumbnail using the first three images let me set the switch to three for the first three images The Prompt will be generate a YouTube thumbnail for a video titled how one guy changed the shoe game in the NBA the following text should be on the thumbnail they all laughed this looks so good maybe not amazing but as far as thumbnails go I'd click on it this could seriously change how a lot of things are done next I want to try creating images and text without giving it an input image I'll bypass the image input by unchecking it I'll go with generate an image of a chubby little anime robot in 80s ANM style wow I'm impressed text to image is amazing but what about just text generation let me go with give me five title ideas for a video on how gemin 2. 0 flash can do anything in comfy UI it only gave me three let's see how good they are the first one is gemin 2.

0 Flash unlash the only workflow you need in comy UI this second is mind-blowing control gemin 2. 0 flash transforms comy UI possibilities and the third is anything you imagine comfy UI can now create thanks to gemin 2.