Leonardo - AI Image Generation & Settings Walkthrough

in this video I'm going to dive deep into Leonardo the AI image generator I'll show you how to easily create images and guide you through the options explaining all the settings such as photo real prompt magic or Alchemy step by step in my last video I showed you how you can join up on Leonardo to use their free plan which lets you use 150 tokens each day if you haven't checked that out you can find the link in the description below so let's go first let's generate an image on the left side you will find all the options on the main screen you'll see a box where you can write your prompt this is where you explain what you want your image to be below the prompt there is a drop down menu with the models you can choose from there are quite a few models in Leonardo like dreamshaper version 7 absolute reality 1.6 or a 3D animation Style there is also the latest stable diffusion extra large model which was released by stable diffusion only a few weeks ago next to that another drop down menu with various Styles these are just additional Aesthetics that you can use to make your image look better by the way most Styles need Alchemy when you turn off the Alchemy slider you're left just with Leonardo and none we'll talk more about alchemy later for now let's keep going on this screen there is a slider to enable a negative prompt if you don't want something in your image like people in a cityscape for example you put it in the negative prompt the image generator takes that into account under the generate button you can see the token count I'm at 20 tokens now because I set the image count to 4. let's change that to one now Leonardo will only use 16 tokens so let's hit generate [Music] I'm really pleased with the results now let's take a closer look once the image comes up you can check it out in the image feed when you hover you get choices you can delete the image download it or copy it to a clipboard then there are some actions that will use up tokens so let's dive into those first off you can zoom out an image that means Leonardo starts painting the edges pulling back on the zoom autofocus this is my original and this is it after unzooming you can also strip the background from your image if you compare before and after I'm pretty impressed with what Leonardo did there are also some cool options for upscaling which means boosting the image's quality so you get a higher resolution image Leonardo has different choices that give different outcomes [Music] there is alchemy smooth upscale and Alchemy crisp upscale by the way if you apply any of these you can switch between the regular image and the upscaled one anytime I've tried them all and here's a side by side from what I can see crisp upscale is closer to the original not really sharper on the eyes small upscale gave me a super detailed high resolution image especially around the eyes and hair but not too sharp to look fake photoreal is the newest thing from Leonardo a powerful tool that helps you create all sorts of images from cinematic photos to Super realistic portraits what's cool about photoreal it's very simple to use and you won't need complex prompts to generate great images however it will cost you tokens what you should be aware of enabling photoreal means disabling prompt magic that's just how it was designed negative prompts won't work either and this feature is currently in beta so it's still under development and testing let's give it a try I can turn on photoreal right here on the left as you can see it automatically switched the model to Leonardo photoreal so I've chosen the Cinematic style and I really like the results [Music] I've tried out a few other style as well here's a comparison [Music] on the left you can spot the added choices that photoreal brings along the first one handles the image's depth of field this decides which parts of the scene are sharp and which parts are blurred at lower values the image has a narrower Focus range with more areas looking soft or out of focus higher values give the image a broader Focus range bringing more of it into sharpness you can choose between low medium and high check out the same image with all three levels for comparison [Music] raw mode is an option that grants you more image control Leonardo suggests leaving it off as it demands more detailed prompts but if you're okay with using more intricate prompts for greater control over your images you can switch it on [Music] at times your prompt may not always deliver the exact image you're after this has been a recurring concern for me especially in stable diffusion Leonardo came up with something called prompt Magic significantly improves how accurately your problems are followed essentially they enhance your prompts in the background to make the images look better you can activate it by toggling The Prompt magic slider on the left there are two versions of prompt magic version 2 the older one when activated it doubles the token usage and there's version 3 a very recent super new version however version 3 is still in beta and can only be used along with Alchemy which means it will cost you more tokens moreover there are extra settings you can apply to prompt Magic the prop magic strength determines how much the prompt magic influences image creation in simpler words it affects how much your prompts are improved a higher number means a stronger effect I'd suggest keeping it at 0.55 turning on high contrast modes leads to moodier images with deeper Shadows if your images appear too dark you can turn this off this contrast option is available only with promptmagic version 2. as with version 3 it becomes part of alchemy and will show up under the Alchemy settings let's give it a try with a new prompt luminous jellyfish drifting in the ocean now let's compare the options when I disable high contrast I get more vibrant colors once I activate Alchemy prompt magic automatically shifts to version 3. the resulting image contains much more detail than in my initial prompt honestly this image here uses prompt magic version 3 with raw mode following Leonardo's recommendation [Music] Alchemy is a feature designed to enhance the quality of your images recently it has been made available for the free tier granting access to a variety of settings to truly understand alchemy's effects I'll begin by generating an image without Alchemy this means photoreal prompt magic and Alchemy are all Switched Off my prompt reads the face of an old man with a white beard and green eyes the image is decent though a bit on the cartoony side now let's activate Alchemy as you can observe it produces an entirely different level of image quality next I'm adding prompt magic version 3 to the mix and it further improves the result here's a side-by-side comparison of my images you can notice that the image quality improves with Alchemy and even more with prompt magic because prompt magic adds bits of detail to the image however it's important to note that this Improvement comes at a cost as Alchemy consumes a significant amount of tokens now let's look into the Alchemy settings high resolution increases the images overall resolution contrast boost manages the distinction in brightness color or tone among the different image elements higher contrast means more noticeable differences resonance influences the image's detail level and how closely it adheres to the prompt Leonardo recommends a value between 13 to 15. [Music] using this slider for public images you can decide whether you want your images to be public or kept private public implies that Leonardo may use them for research or marketing however this Choice isn't accessible in the free tier you'd need to go for a paid option in the image Dimension section you can set your image size by picking one of the preset options or by fine-tuning the width and height sliders you'll see that the aspect ratio shifts along with the size changes alternatively you can modify the aspect ratio directly [Music] the guidance scale is a setting that determines how closely Leonardo should follow the instructions you provide in your text prompt when the scale is lower Leonardo gets more creative when it's higher it adheres more closely to your prompt however this doesn't always guarantee the best images the default setting of 7 is usually sufficient let's compare the guidance scale using the image prompt for a cute panda at the lowest setting it generates an image of a girl wearing cat ears so I don't see a panda here at the highest setting it gives me what I want a cute panda interestingly the recommended setting of 7 produces the exact same result [Music] in my upcoming video I'll be walking you through some Advanced functions like prompt Generation image to image creation and control net I'll also guide you through the possibilities in AI canvas be sure to subscribe so you won't miss out thank you so much for watching and see you next time

Channel: TechXplainator

Views: 2,777

Keywords: ai, ai revolution, ai art, ai art generation, ai art generator, ai image, ai images, ai image generator, ai image generation, text to image, ai text to image, leonardo, leonardo ai, leonardo.ai, settings, leonardo settings, unzoon, upscale, remove background, background, photoreal, photo real, prompt magic, alchemy, public images, guidance scale, tutorial, leonardo tutorial, leonardo ai tutorial

Id: rVckz2SyaYk

Length: 12min 0sec (720 seconds)

Published: Tue Aug 15 2023