Additional Notes

Note: This section is a work in progress and needs some refinement.

Those who are familiar with or are practitioners of using Keywords for better targeting and optimizing your content, the below may seem familiar. When we describe what type of image we want in the finest possible detail, we will start getting better and better results as we begin to make our description more and more elaborate. The text input one provides essentially forms a set of keywords. These are inputs which you will use to communicate with the AI based designing tool.

Bear in mind that the AI imaging tools are incredibly smart, but they are under active development and they may not be able to understand the command in adequate detail. Just the way when we speak in a language or utter certain words, which may or may not be understood by the person on the other side. Similarly, the underlying technology behind different drawing tools such as Mid Journey, Stable Diffusion, and Dalle2, all likely use different interpreters at the back-end. Therefore, the results may vary for different types of image creation tools.

For the purpose of this segment, I created two sets of images with identical descriptions on different tools or sites, the websites that I used, were a night cafe, the one was teamsnap. Third one was a super machine dot app for which I got the paid subscription by AppSumo. Then there were a couple of others that that I also used, and the output that I got was different for each, the moment I changed the form factor from landscape to portrait, to portrait or to square, the image output still look the same, you know, colors still look the same. So for example, let's say it's a picture of a woman wearing a purple sorry, so, it was still a woman wearing a purple sari and each of the images, but then they were not similar, you know, the person's face had changed the person herself had probably changed in this case and so on. Versus if we use the seed ID, the image that is created on a similar tool irrespective of where you create it.

So, if you any, if there are three sites that all three of them use stable diffusion as the backend and you give them the seed number, they would more often than not generate very similar if not identical images. So, what we learned here, number one was seed ID second is volumetric lighting, third is what is our germ fourth is what is orthographic projection, what is 8k imaging? What is highly detailed, what about high resolution, so, he spoke about eight key high resolution is also similar for that in certain cases, there are sites which may not allow you to use certain band keywords. So, you know, children related jazz names such as you know, little boy, little girl, youngster and so on children's bedtime story.

So, everything under the children umbrella may not be you know, it will not get accepted as a command a night Cafe does that this one does that stable division which is hosted on hugging phase has that filter there are sites which may have profanity filter, so you cannot use swear words no Gore, you know, no violence, no sexual violence, etc, etc. Third is there are sites which may block Not Safe For Work keywords. Bear that in mind. And just yesterday, I had a very positive experience with one of these image creation sites. It's a family friendly images and their pitches also that we do family friendly work. I purposefully used some creative keywords to you know, that mean render images, which may be on the ACR site for lack of a better term, and immediately their system flagged it and displayed an arrow to the developers Congress.

lating them saying that you know, some artists some people may try to take a creative approach by using a different language by using a different set of combination of keywords etc. And it may result in images which are either in you know, trying to outlive their create their fantasies or you know, fetishes in nude form of any celebrity and so on is a classic that I have seen across the board, but to prevent that there may be these keywords in place.

So, this is this is a sort of the very basics of it in terms of materials, different materials or you may have oil on canvas ink dot art and so on. Then you have etching sketching busts, you know marble statues, you may have all these different formats, you may have stick image, those are actually fun, you will have charcoal drawing in terms of material continuing on that we have the pencil sketches, we have pen and pencil and then when we are talking about level of detail and that is where it really gets out of hand sometimes for some people and that will be how much level of detail do you need. Now, you know, very detail highly detailed volumetric lighting, sunlight, sun is shining, golden hour Golden Ratio, these are all the terms which if you're not familiar with the space of either photography, and I'm not an I'm learning sail, or if you're not familiar with the arts in general, it may get very confusing and overwhelming for you.

So in this segment, I'll try and simplify some of that for you. And maybe there are about eight or 10 different keywords and five or six different drawing you know design styles or designers themselves that probably would contribute to about 80% of the images that you may create or you may see in this space. That was the intention that once you have a fair idea of what what days what this particular artists or particular style or or image or layer or pattern stands for. You may start experimenting and you can also then go ahead and contribute back to this the exciting space of AI based image generation with that wrapping up this glossary segment thank you so much.

© by Amar Vyas, 2022 - 2023. All Rights Reserved. Built with Typemill.