AI Generated Image No. 673
Made with the Bing Image Creator, powered by DALL-E 3.
When I make these images, I start with a creative concept in my head, and then if the output is good I will start refining the text prompt, or start adding more complexity to it. It is vital to learn something from EVERY single image creation attempt about what the AI model does and does not do well.
I find that using AI image tools to try and create a very specific image, and trying to use it like a 3D art program, does not work well.
I approach this work like photography, where there is a wide range of acceptable outcomes for the image. Like with photography, the machine handles the technical side of creating the image, and the human is in charge of curating/directing the process.
If I go out and shoot photos, there are almost infinite images that could be created, but I am only going to be taking perhaps 200-500 shots.
It is up to the photographer to curate this process to decide what images will be taken, with the goal of maximizing the probabilities of something good being created.
The same concept applies to AI image generation, which I think is best compared to photography in terms of the skills involved. The machine handles the technical parts of creating the image, but the human is in charge of curating the process in order to maximize the probabilities of something good being created.
One more thing is that these are done with the Bing Image Creator, and I would not be surprised if it gets shut down completely or severely restricted in terms of what you can do with it.
People are pushing past the content controls it has, and I would not be surprised if Microsoft just shuts the whole thing down at some point, but hopefully that is not the case. They could easily decide that the liability from this is greater than the benefits.
Also, it must cost a fortune to create all these images, in terms of the computing power required to offer this service for millions of users.
- Josh
AI Generated Image No. 673
Made with the Bing Image Creator, powered by DALL-E 3.
When I make these images, I start with a creative concept in my head, and then if the output is good I will start refining the text prompt, or start adding more complexity to it. It is vital to learn something from EVERY single image creation attempt about what the AI model does and does not do well.
I find that using AI image tools to try and create a very specific image, and trying to use it like a 3D art program, does not work well.
I approach this work like photography, where there is a wide range of acceptable outcomes for the image. Like with photography, the machine handles the technical side of creating the image, and the human is in charge of curating/directing the process.
If I go out and shoot photos, there are almost infinite images that could be created, but I am only going to be taking perhaps 200-500 shots.
It is up to the photographer to curate this process to decide what images will be taken, with the goal of maximizing the probabilities of something good being created.
The same concept applies to AI image generation, which I think is best compared to photography in terms of the skills involved. The machine handles the technical parts of creating the image, but the human is in charge of curating the process in order to maximize the probabilities of something good being created.
One more thing is that these are done with the Bing Image Creator, and I would not be surprised if it gets shut down completely or severely restricted in terms of what you can do with it.
People are pushing past the content controls it has, and I would not be surprised if Microsoft just shuts the whole thing down at some point, but hopefully that is not the case. They could easily decide that the liability from this is greater than the benefits.
Also, it must cost a fortune to create all these images, in terms of the computing power required to offer this service for millions of users.
- Josh