Learn DALL·E 2 in 20 MINUTES! DALL-E 2 Beginner Tutorial

Key Takeaways

hey friends welcome to one little coder in this video we're going to learn how to use dali so this is your simple tutorial on how to use open hair is dali if you get dali access so what are we going to cover in this video first we are going to learn how can you create a new image using dali or if you also have an image how can you use that image to make some edits or create some variations of that image so dali has different kinds of input and we're going to explore all these different aspects in this video so at the end of this video you'll be able to fairly understand different features or especially i should say different features of dali interface that you would see on the dali web app um it is not going to cover tips and tricks or features of dali in itself so this is purely helping you get started with how to use dali so first thing is when you log in like let's say you you've got an access email so the access email would look like this and once you click then you would log in with your id and then once you log in with your id you would land on a page landing page like this so in this page you've got a couple of options the first option is dali in itself is a text to image model which means you can give a prompt a text prompt natural language just like this here and then it's going to create an appropriate image for you it's going to generate an image that is one type of input the second type of input is you can actually upload an image to dali and then ask dali to do two things either you can ask dali to edit that image or you can ask dolly to create variations of that image so now we're going to start with the text prompt before even you could you know give a text prompt because giving text prompt itself is something you know called prompt engineering these days so you can click surprise me and then dali would itself put a prompt here and then you can click generate and then see how the response looks like while the generator response is while the image is being generated you could also see certain tips and tricks from opena themselves to help you create better dali images so the detail like as much as detailed as much as detail you can incorporate in your description the prompt your image would be more detailed because it's going to basically use the text that you give and then try to fit it and you can see it says longer specific descriptions tend to work best and we're going to just quickly see that so this is what a bowl of soup that looks like a monster knitted out of old so that's what this is so now what i'm going to do is i'm going to click the dolly icon again so that it will take me to the home page and you can see certain images like examples here so you can see certain examples if you want you can click one of the examples and then that would give you one of the important thing when you get access for the first time is you need to remember that for one day for 24 hours time period you get to generate 50 prompts you get to try 50 items which means um in every prompt you have six images i think it's a grade of six so which means you can generate six uh if you say six into fifty so which is approximately 300 images but if you just purely look at from a prompt perspective you can try out only 50 prompts um do not frantically try everything like that's what i did when i got my first access and then i think i ran out of my access for the next 20 hours just like in a couple of hours so make sure that you want to try out what all those things and then you can use it another thing that i did at the start is like i had a google sheet where i used to enter the prompt that i'm trying and this also helped me in understanding the prompts that are violating open ais policy so there are certain prompts that open a would restrict you from trying for example if there is a violent word like fight gun so these kind of prompts would put you like flag your query and if you try out a lot of prompts where your query or the prompt is being flagged by open a there is a very good chance that your account might get banned permanently or temporarily so make sure that you don't you don't do those things so let me start with a very simple example description um this is a prompt if you're if you have never heard of prompt prompt is like the starting point which you are telling the model to use as a queue to create something in text generation model prompts will generate new text in image generation model prompts will generate new images so i'm going to say a d a 3d render of microsoft microsoft clippy on a computer disk running windows 95 let's see what is going to happen so what are the things that i'm trying to do first i'm trying to tell dali that i want a 3d render next i want a 3d render of microsoft clippy and i want this clippy to be on a computer disk now i am assuming that dali would understand that the computer there is running windows 95 and and this is this is what i have gotten result and i would say this is not this is not a bad result given that what what i expected so this like let's let's move on from this and then try out something completely different so i'm going to say dali a nasa picture a high resolution nasa picture of a detailed um nasa picture of our galaxy captured by an infrared telescope so again what i'm trying to tell i'm trying to tell dali that first i want a high resolution picture it's not going to change the resolution of the image in itself it means it needs more detail more like sharp images i'm saying nasa picture so the dali knows like we are trying to explore something like nasa then we are saying that we want a picture of our galaxy and we want captured by an infrared telescope and and honestly these are really good pictures these are pictures almost as similar as what nasa released a couple of days back and that was captured by um jwst james webb the telescope so these are really good i'm going to try out one final um prompt but you can understand the way i'm building from the reason i'm trying out different option is for you to get an understanding about how we are trying to prompt so i'm going to say um a minion um a minion sitting on a fast moving car fast moving yellow cab in front of a golden gate bridge so a dslr pick so i want a ds dslr photo of minion sitting on a moving hello cab in front of golden gate bridge again you can see the details so the dslr photo specifies the quality of the image what do i want i want a minion but i want the minion sitting on a fast moving cap so when i say fast moving the reason i want it is because i'm using a dslr effect i want some kind of you know difference in the way it looks and i wanted to be near or in front of golden gate bridge and i wouldn't say i've got the best of the images but you can see fast moving because i said fast moving how it looks and what kind of images that we have got so this is this is basically how you give a text description but like you said as detailed as you can describe what you want an image you can get better outcome so that is that is one thing about giving a text and then asked asking dali to create something out of it so i'm going to go to the second section now first text section is finished in the second section you can upload an image and then you can ask dhali to create variations of the image or you can edit it for example i've uploaded this image this is an image that was released by nasa a couple of days back i mean i think yesterday and once again just to show click upload an image select the image that you want to upload and then say done you have to crop it and once you reach the screen you would get two options edit image generate variations editing image will help you edit parts of the image or in fact even change the whole image if you want to add something if you want to delete something if you want to do in painting if you want to do extension so all these things could be done with edit generate variations will create different aspects like whatever dali understands from this image based on that understanding the ali is going to create new images so first i'm going to click generate variations so that we can understand what kind of variations come in and after the next section we can move on to the edit image part and as this is loading you can also see that there is there are you know certain tips always pay attention to these tips you know you might find it repetitive but you might get some really good idea when you pay attention to these tips so now once you upload this you can see that this is your original image that you uploaded and this is these are like the variations of that image particular image that dali has created and i would say this is really impressive so this is one way of creating variations so the other way of creating variation is for example one of the things that we tried for example i i can say a large dolphin a swarm of dolphins swimming across atlantic ocean captured by a scuba driver okay so now this is this is an im prompt i've given and it's going to generate image so what we are going to do is we are going to pick one of the images and then ask dolly to create variation so in the previous example what we saw is we uploaded image ourselves and asked dali to create variations of that image here we are going to select one of these images for example i can select this image or i can select this image and i can tell dali ok dali now i want variations of this image and i can edit this as well and i can create variations so click variations now this is going to create set of images so now every time you click something edit or variation or new prompt it's a call so in one day you can make 50 calls just to remind you um in 24 hours time period so just just just use it wisely i mean when you get access you would be very tempted to use all 50 um one or two days but then later on it might change at least it happened with me so this is an original image but not created by us that's actually created by dali but then we use that image to create further more images and and that's that's how variations work i can give you one more example i can say uh create like a high qual carton a a cartoon picture of miss marvel so now dali is going to help me understand design or help me design couple of pictures of miss marvel now what i can do is i can ask dali to pick one of the images and then ask dhali to create variations so the first feature that we saw was creating image from text now the next feature that we are seeing is creating creating using prompt to create an image and then use that to create new variation so i've clicked that and then i'm using that to click create new variations and variations could be done on either the image that dalia had created or the image that you are uploading one thing that you need to keep in mind is you need to mostly keep everything like um non nfsw not suitable for non s in sfw so it shouldn't be not suitable for content like it's like you you cannot you cannot do like create order content or anything like that the second thing is that you need to keep in mind this when you are going to upload image you cannot upload images with human faces even if it is yours so what happened this we asked dali to create a miss marvel cartoon picture and we picked one of the image and then we generated more variations of it and you can see the variations of it it would it would look almost similar like um that's what the whole point of variations is now we learned about text to image then we learned about variation now i'm going to show you about a editing option so what i'm going to do is i'm going to go to home page again click upload an image select the same same galaxy image and done and now i'm going to click edit image when i say edit image i can select a particular area that i want to edit and you can see the brush size and i can adjust the brush brush size here like you can see my size of the brush going up and down here so you can see like it is small now it is super big now i can i can do all these things so um very small and super big for example let's say i want to put a planet here okay this is a place i want to add a planet so i can say a beautiful blue color planet and i can say generate now what dali is going to try to do is dhali is going to try to find that area like that particular area and create an image that goes well with the background image like the image that it has got and it's going to fill in with a very beautiful blue color planet i i'm not sure what is it whether it's going to be beautiful or no but uh let's see so did it do anything okay i don't see a planet okay maybe it's trying to create a planet here but i think it didn't fit very well so what we can do is we can now try different prompts like still we are in the edit mode but it is going to try to fit in in the same place so what we can alternatively do is we can upload an image okay and then create a much bigger space like for example i can select all these place and then say a planet like saturn with rings and some moons a photo fill in with the planet okay let's see if it if it can if it can fill in so like i said so the mask that you create using the brush is the area that dali is going to focus on trying to create based on your prompt text prompted that uk and then it's going to try to fit in with the original image that you uploaded and this is called image editing feature in dali image editing could be used for a lot of things um i mean it's very cartoonish uh it looks silly but you know you can see that how it fits and how it blends it like even you can see the images blending in and all those things so so this is this is how that particular feature is like for example um i can upload an image and i can say okay edit the image i can say this is a place this is the place where i want fill in with a sad programmer waiting for the code to run in front of a laptop i'm not sure again whether it will work because you know context also matters like background you've got an image of the galaxy and in the front you are asking dali to create a programmer and then fit in there but the whole point of this exercise like i think the reason why open air has given us access is to try out all the wild imaginary things that we could come up with whether it is art whether it is different kind of prompt you know what so so just just try it out this is the original image and you can you can actually see a programmer sitting there uh i don't know why unicorn or something is there and there is uh there's a depressed disappointed programmer sitting there but i think you get the point so this is how you edit an image so so far what we have seen is first we saw to give an empty text and then create an image second we saw creating variations from the uploaded image third we saw we can edit the image from the image that we uploaded finally before closing i'm going to show you how you can pick an image that is like generated by dali and then you can edit that image as well like for example this is a prompt where we have set a 3d render of an astronaut walking in a green desert and this is one of the examples that they've given and we're going to pick that image okay so we have this all these things and we are going to pick one and then we can say edit and then i can basically say um i can say um fill in with a robot waiting for baggage in the waiting in the airport i don't know if waiting in the airport this is right thing because given the context again like you have different context there um but let's see what happens so now what we are doing is we actually generated image using dali first using text form and using that image we are trying to make edits so it's not complete variation so it's like um so so you have this and then you know like robot waiting in the airport so so you know there is there is no robot but you can see um like waiting in the airport thing so so this is this is how you can pick an image that is generated by dali and then you can edit it and now i think at this point you know what to surprise me is what does upload an image what is generated and all these things so the next thing that i would like to show is when you generate daily image reproducing it is usually difficult like for example we recently did something you can go to your history and then see but if you do not have it in history then it is very hard for you to recreate it or get it out back so for example the prompt is a photo of a white fur monster standing in a purple room a photo of a white fur monster standing monster what did i spell monster monster standing in a green color light room okay so this is this is my prompt you can see i've copied my prompt and you can see the images that are generated now what i'm going to do is i'm going to generate the same set of images again and and you can see like how similar like if you get the same images again i think it's quite important when you have detailed uh prompts sometimes you want to generate the same thing so you have you have this now when i click generate again you wouldn't necessarily see the same image for example let me like for the next thing i'll copy a different one but uh i've just used the same exact prompt and then we have tried to create um a different image i mean like a different run different call we are making a new run but using the same prompt and you would see that not very surprisingly you have totally different images now if it has recent you can go to the right sidebar and then get that image but if it is not recent then you cannot do it so for that what you can do is you can like for example if you generate something um like a synth wave themed smart phone is what you are generating then what you can actually do is you can save the image and that image will become part of your collection so now this is almost the last section of this video where we have already learned how to create images now i'm going to show you how to save that image in your collection so that you can use later so this is this is a synth wave themed um smartphone so for example let's say this is my image and or this is my image whatever that image is this i can click this button so this one will download the image i can click this button and this will report to dali if there is anything like open a if there is anything wrong on the image like let's say you find bias or anything or i can click save once i click save it goes to my collection and i can click my collection here and then you would see you would ideally see this image so i've got one image so this way you get to create the same image you get to create like you get to see the same image again um without losing the image so this is um i mean you don't have an option to set a seed value here but you can save the image and then get the image back and just like that once you go back to your business then that's gone so if you want go to your collection and then you can find the image i think these are the things that i wanted to cover like to quickly summarize first we gave a text prompt and then designed an image then we tried a different variations of text prompt then we learned how to upload the image and then we get variations then we learned how to use the text from create an image and create variations then we did the same thing with edit upload an image and did it and create an image with dalia underneath it and then finally we learned how to save the image i hope this helps you in speeding up your dolly journey um i mean like if you have got access to this tool you should be thinking you should be spending time on creativity not figuring out how to use this tool that's the whole point of this video and i hope i've done justice in that and if you have any questions let me know in the comment section otherwise happy coding happy darling peace