[Fair warning – I’m cheap, my website is slow and this is very picture intensive. It will load slowly]
When I started Opinionbypen.com, I quickly learned that people wanted pictures. There’s a number of photo sites out there but between usage rights and finding the right picture, I quickly turned to Poser as a way of illustrating my blog posts. I was surprised to find I also enjoyed the ability to create images based from my imagination. The picture on the left is one of those. Even so, it’s the picture on the right that I’ll be discussing today. That picture was generated using DALL-E, an AI image generator.
DALL-E is provided by OpenAi and like ChatGPT, you type in a description of the picture you want and it draws its interpretation of what you want. Actually, it draws four images and allows you to generate variations of the one you like the best.
It’s not without frustration. After several tries, the geckoes with the Happy Holidays flag were the closest I could get to duplicating my 3D render of geckoes using semaphore flags to wish you Happy Holidays.
Like ChatGPT, the more detailed the description, the better the results. However, a little ambiguity may produce something totally unexpected and even better for your purposes. I admit I found DALL-E intriguing. I managed to generate a picture suggesting my dog was cheating at cards (fortunately Oliver has a common face but the eyes are wrong) and a picture of the lead character in one of my unpublished novels.
Novelty aside, why would I want to use DALL-E? It’s hard to get exactly what you want and its human renditions are straight out of an alien takeover. More about that later.
With Poser, once you have a scene built, it’s easy to modify. There may be a way to do that with Dall-E but I haven’t found it.
While writing this, I realized that I enjoyed the creation of my wizard. I wanted to viewer to wonder where he’s going, why he’s in a frozen wilderness and how his plane can stay in the air? The picture created by Dall-E is simply an image and evokes none of those questions.
The people created by Poser look remarkably lifelike. I believe the DALL-E limitation is deliberate. When I asked for a high quality picture of a man in DALL-E, I was warned that I was violating their standards.
When you are looking for a way to illustrate a point, Dall-E can be a great choice. My average Poser picture can take six hours or more and Dall-E takes about two minutes for each attempt. I can’t get exactly what I want but often I can get close enough.
As long as it doesn’t include a human, DALL-E can produce remarkably realistic images.
You can change the style and tone of the picture with just a few words. As an experiment, I provided a description of my favorite YouTube streaming site and asked for several different styles. I added those pictures to the end of this post.
DALL-E is much cheaper than Poser. Your first 50 DALL-E images are free. After that you spend $15 for 115 pictures. I burned through 100 images just for this post. On the other hand, I spend almost $15 for each Poser model I buy. Once bought, I can reuse the models as often as I want but I still end up buying one or two models for each Poser picture I create.
Even if you don’t own a website or need to illustrate a book, DALL-E fun to play with. Especially for the first 50 free images. I’m not giving up Poser but you will see me using some DALL-E images in the future.
All these images were produced using the same words. I only changed the style.
© 2023, Byron Seastrunk. All rights reserved.
Another great post Boss