Why CGDream AI is Superb for Art Generation from Text, Images, and 3D Models

hi in today's video I want to talk to you about an AI that can do text to image image to image photo realistic images paintings 3D render looks and mockups what's unique is the 3D model to image feature that's right you can use a 3D model rotate it and then generate an image from it uh you can go to the CG dream AI website and test it for free um if you like it you can uh switch to the premium version to remove limitations right when you open the website you can see what other people people have generated from text prompts uh you can also see Generations based on images if you hover over an image you'll notice a small icon uh indicating it used a reference image uh the third option and my favorite is uh Generations from a 3D model these are marked with a tiny Cube icon to show um they were based on a 3D model you can sign in with your Google and Facebook or CG Trader account once you are logged in you can start experimenting describe your prompt with details about what you want to generate including the subject details environment lighting style and so on you have two options you can generate the prompt as it is or you can use AI to enhance your prompt this way even short prompts will result in nice Generations I will use the generate option in this case depending on how many people are using it you might have to wait for your generation sometimes it takes a few seconds and other times it might take up to a minute I'll speed up the video when I edit this recording so you don't have to wait when you um hover over the image you can see some buttons the down arrow button lets you download the image next to it there's a a hard icon that uh lets you add the image to your favorites if you click the favorite menu um you can find all your favorite images there you can also delete your generation from this icon if you click on the 2x button you can upscale the image by two times but you will need to wait a little longer for it to be upscaled once it's upscaled the 2x button moves from the right to the left side allowing you to switch between the original and upscaled versions as you can see the upscaled image is clearer and sharper you can also submit your images to challenges click on the challenge menu to see what challenges are available then click on rules and prizes to check the timer see how you can participate learn how to submit and find out the categories and what you can earn click on the my images menu to go back to your Generations let's try another prompt cute cartoon bunny in a magical forest on the right you have more options um you can choose the aspect ratio the first is square the second is Portrait and the last two are landscape orientation you can generate from one to four variations at once prompt guidance Works similarly to the CFG option in stable diffusion you can Auto enhance prompts is using dreamup and ads related filters to prompts that have three or fewer words the seed can be random or fixed and you can also add negative prompts for things you don't want to appear in your generation this time since the prompt is short I will let the AI improve it and not bad for a short prompt but we can do even better next to the prompt area you have filters you can also press the enter key shortcut to call up and update filters and you can add up to five filters these are Laura models trained in a certain style filters are IM immediately recommended based on the entered prompt but the filter search overrides The Prompt recommendation allowing you to search for a different filter and you can pick from a list of over 200 filters or search for your favorite um let's see what we can find with u fantasy and I think I'll go with fantasy art and um as you can see it's added to the left let's search for another one something with cartoon and I'll choose this manga cartoon now let's find another one this one with dressed animals looks interesting um as you can see I now have all three filters and uh you can use the sliders to tell the AI how strongly each style should influence the generation you can also delete them using the recycle bin icon hit generate when you are ready um I really like the result it's nice and clean um if you have premium you can turn on the private toggle for future Generations um you can also check each image for a lock icon which you can click to make the image um private or public the images created with filters have a lines icon in the bottom left I will add it to the favorites really quick um here you can see a few of my favorite Generations let's go back and one thing to remember is that if you want to do something different remove the existing filters or uploads from the left now that we've talked bit about uh text to image Let's uh switch to image to image and 3D model to image you can upload an image that you'd like to use as a reference for example um I have this cute bear now there is a slider um for resemblance that controls how similar the generation should be to the original photo I'll try with 0.35 to see what I can get and for the prompt I'll write something like cute cartoon panda and wearing a a ballerina addess I'll add a white background um and for style maybe digital painting and Kawaii after hitting generate and checking the result I got a nice generation but the original bear is too strong so I'll reduce the resemblance and and try again the panda is starting to appear so we are getting closer I'll reduce the resemblance even more it's not a standard value and depends on the image so you just have to play around and and see what works for you now we have the cute panda as I wanted I started with this bear and ended with this cute panda another thing I like is the in painting option if you use this brush you can paint over the area where you want to change something for example I want to add a gift there now as you can see in the prompt area it says in painting so you put the things you want to appear in that area in my case a pink gift box hit in paint button and uh the result is um a panda with a gift box now let's try a 3D model but first remember to you know delete the image reference click upload and uh choose a 3D model with the extension fbx orgb I have a Halloween coffin I recently made in blender uh you can rotate and move the 3D model into the position you want if you hold the left Mouse button you can rotate around with the right Mouse button you can pan or move the object and with the mouse wheel you can zoom in and out it's pretty simple if you choose the walk option you can navigate like in games using the W as and D Keys however for 3D I prefer the default orbit option the walkthrough option is useful if using an interior or environment 3D model as orbiting is inconvenient in those cases orbit is useful when there is a single object in the center I think I will go with the front view and you also have an option to untoggle the ground if you want the object to appear in the air I'll quickly describe in the prompt uh how I want the coffin to look including details about the environment the material and so on then I'll hit generate and this is the result pretty cool if you see an image with this um Cube icon in the corner you can click on it to see the 3D model used to generate that image in my case it was the coffin let's move it around and maybe get a close-up shot of it this time I'll play around with some filters maybe uh all maybe I'll choose a material um and cartoon filter uh change the prompt and then hit generate sometimes you have to click on my images to see your generation at the time of this recording uh it's still in beta from what I heard so it might have some bugs while I wait I can upscale the previous generation I'll skip to the result which is this one pretty interesting now I will um remove the filters and the 3D model if you don't have a 3D model you can choose from existing ones in the 3D model Gallery it has all kinds of 3D models to choose from CG dream is part of CG Trader which you might have heard of if you've worked with 3D models and done uh 3D renders I will choose this perfume bottle it doesn't matter if it has labels or text since when you add the prompt you can make it look how you want let's say um I want to generate a mockup for a perfume bottle maybe I'll place the bottle on the left and uh leave some space for text on the right I'll search for a filter maybe this um you know product ad filter will work well for what I want then I need to describe the prompt I'll add it quickly here and hit generate I misspelled some words so I'll adust adust The Prompt quickly and try again now I got exactly what I had in mind um you know without a 3D model it would be um hard to get the exact position of the model um I'll click on the cube icon and try a different angle U with the same prompt um to see if I get a similar design um the result is quite similar um if you want almost the same design you probably need to use the same seed and mention more details like how the bottle cap should look and so on you can generate with an empty prompt if you are you're using dreamup just click on it and it will create a random generation for example I got this cute kitchen interior auto enhance uh for short prompts is uh on by default let me turn it off and test it um without this feature to compare I will try a short three-word prompt like a white bunny uh since we didn't provide more information we get a normal bunny let's get more creative by adding filters I'll add dressed animals also plant grass could be interesting I also want a cartoon one so I'll search for it in the filters and pick manga cartoon hit generate remember that you can change the filter influence by moving sliders you can delete filters or add up to five filters in total and the result is this cartoon bunny let's turn the auto enhance for short prompts back on to see how it influences Our Generation click generate and since the prompt is three words or fewer it should automatically use dream up and filters the result is this clean cartoon bunny dressed up in the grass so we started with a basic bunny then added filters and finally used both filters and auto enhance uh with AI for the last version if you want more control over the generation you can add more details to your prompt for example um you could mention that the bunny has glasses is in a magical forest um includes some colors and um add a lamp then hit the dream up button you can generate multiple images simultaneously hit generate second time and then once more for the third time another option is to use four variations instead of one to get more images and here are the results I think they came out pretty cute what do you think I played around with the same prompt but this time I added a few more filters like cartoon fantasy and cartoon animals um I think it turned out quite cool so I will submit it to the challenge using this button uh I just got a confirmation that it was submitted you can also find ideas for prompts or remix other people's prompts by going to the gallery uh there you you'll see a lot of public Generations if you click on one you can see uh The Prompt and the filters used for that generation let's go back and see what else we can find I like um this woman portrait so I clicked on it uh on the left uh you can see it has uh two filters and this is the prompt used if you want something with the same um prompt but different variations uh you can change the filters for example I can add witch and uh fantasywoman filters and um maybe reduce the metallic filter um the result is this cool portrait I can increase the image resolution by um upscaling it uh you can see the original image and uh the new upscaled image I always add my own twist to The Prompt um to make it more interesting so um instead of a woman I put a witch and instead of head phones I added a witch hat I removed the part related to the headphones and instead of listening to music maybe she should be casting a spell I hit generate and this is the result I got how cool is that if you need help you can click on the community button that leads to a Discord server where you can post um your questions in conclusion what I like the most is that I can use 3D models to get more control over the generation I hope they add even more models in the future um and provide more control like the ability to add multiple 3D models in a scene the filters or Laura models are quite useful and there's a big variety I appreciate that they offer a free version so people can still use and test it without paying but the premium option is quite cheap compared to Mid Journey or Dolly uh I still sometimes do 3D renders but it is very timeconsuming to uh set up materials lights and you know wait for the render uh being able to use a 3D model and get a quick render with only a few words is quite useful I am trying to make videos on AI tools that can help the community I had to turn down a few collaborations because they tried to dictate what I should say like claiming their product is the best on the planet instead of allowing me to do an honest review of their product so no um I don't do things like that uh leave a like if you enjoyed the video thank you for watching um and have a great day I want to thank CG dream AI for sponsoring this video [Music]

Channel: pixaroma

Views: 2,559

Keywords: CGDream AI, CGDream, AI art generation, text to image AI, image to image AI, 3D model to image AI, free ai art generator, AI art tool, AI art tutorial, how to generate art with AI, art creation with AI, AI filters for art, CGDream AI tutorial, AI design tools, art mockups with AI, digital art tools, CGDream AI features, AI art software, creative AI tools, artificial intelligence, pixaroma, generation from 3d model, free ai tools, ai tools, ai art, free image generator ai

Id: 7U-7UUopUSE

Length: 15min 16sec (916 seconds)

Published: Mon Jun 10 2024