2 things: First up, Clearly I have a LOT to learn about prompting. That prompt is not even just 'next level'; it's another language. I gave characters one try in Flow, couldn't figure it out and just carried on doing what i was doing. I will continue the research...
Secondly - who you calling old?! BTW, I also remember the excitement of the 56k modem bounce waiting to see what new 'treasures' had been added to the cornucopia!
AI Vids
-
Chevy Monza
- Posts: 10
- Joined: Fri Aug 14, 2026 8:22 pm
Re: AI Vids
Firstly, I wanted to add to the cheers, these videos are next level; appreciate you sharing them here.
Secondly, Its great to have a place where not only is there active sharing going on, but some really good discussions and sharing ideas of how to prompt; and the tools out there.
Secondly, Its great to have a place where not only is there active sharing going on, but some really good discussions and sharing ideas of how to prompt; and the tools out there.
Re: AI Vids
Happy to help. I appreciate the compliments on prompting--it is a challenging art to learn. But here's the thing:
Google Gemini taught me how to prompt.
I certainly didn't invent that "Shot with 5-perf IMAX 70mm footage, Atlas Lens Co. Spherical IMAX 55mm T3.5., Subsurface Scattering (SSS), Micro-contrast, Uncompressed 16-bit RAW, CNC Machined Tolerances" technobabble on my own. But it works because it tells Gemini "how" the image should look by referencing a specific camera, a specific lens, and a specific film. Prompts like "Super Detailed 8K" are meaningless because there is no such thing as "8K." Note: Prompts don't work the same on all platforms. But the one I provided should work on the more advanced ones. The new "ARGON" setting in Gemini is great at making images, and you can also set up "GEMS" with all the references images locked in.
If you use Gemini, it can teach you how to prompt too. The text chat feature on Gemini is fundamentally different from the image generation feature on Gemini. So the text chat feature in Gemini will show you how prompt the image feature.
Just keep in mind that it's not overnight. There is a learning curve and there is a lot of trial and error.
Google Gemini taught me how to prompt.
I certainly didn't invent that "Shot with 5-perf IMAX 70mm footage, Atlas Lens Co. Spherical IMAX 55mm T3.5., Subsurface Scattering (SSS), Micro-contrast, Uncompressed 16-bit RAW, CNC Machined Tolerances" technobabble on my own. But it works because it tells Gemini "how" the image should look by referencing a specific camera, a specific lens, and a specific film. Prompts like "Super Detailed 8K" are meaningless because there is no such thing as "8K." Note: Prompts don't work the same on all platforms. But the one I provided should work on the more advanced ones. The new "ARGON" setting in Gemini is great at making images, and you can also set up "GEMS" with all the references images locked in.
If you use Gemini, it can teach you how to prompt too. The text chat feature on Gemini is fundamentally different from the image generation feature on Gemini. So the text chat feature in Gemini will show you how prompt the image feature.
Just keep in mind that it's not overnight. There is a learning curve and there is a lot of trial and error.
Re: AI Vids
To be clear, "8K" exists as a video format, but nothing you and I are using right now for AI image generation is gonna make 8K images. That aside, using a generic "8K" prompt is the kind of generic, non-specific prompting that AI will almost always disregard. AI image generation platforms will be "lazy" if the user doesn't pin it down with solid, streamlined prompting.
When I was first learning, I think I asked somebody on this thread for some pointers. They were very generous in their advice. And summed it up by saying that AI image generation is like a "slot machine." And it is. But solid image prompting stacks the odds in your favor.
Video generation used to be the same way because you were essentially asking the AI to take a single image and deduce, i.e. guess, what happens next. Fortunately, if you use video generation platforms like Sea Dance 2.0 or 2.5 in CapCut, you can string several images together so that the video prompt can be fairly simple. That is because the images themselves tell the AI video generation platform EXACTLY what the video is supposed to look like. For SeaDance 2.0, you can string together up to 9 nine images to make videos up to 15 seconds in 1080 or 4K video format. In SeaDance 2.5 in 2.5 you can string together up to 30 images for videos lasting up to 30 seconds in 1080 format. The example videos I uploaded were made using 6 images strung together.
I know it may sound like I'm hyping for Google or SeaDance or CapCut like a paid promoter or high net worth corporate shareholder. (I'm not). I just find them to be what works for me right now. If somebody has different methods or platforms or ideas, I would genuinely love to hear them. And we're VERY fortunate to have a forum in which we can share ideas and methods. Again, kudos to Barbarella for setting all this up.
I understand and appreciate that $$$ is a real limiting factor for all of us, including me. This stuff ain't cheap, particularly if you're using the more advanced platforms. But, IMHO, spending the extra money for better tools, ie, advanced software, hardware, etc., saves you money in the long run because you're not generating AI content over and over, hoping for HAPPY accidents.
If we help each other by sharing information, ideas, good prompts, etc.,we will ALL save money for each other.
That's why I'm being so verbose in these posts. I want to help everybody who is interested IMPROVE at this.
When I was first learning, I think I asked somebody on this thread for some pointers. They were very generous in their advice. And summed it up by saying that AI image generation is like a "slot machine." And it is. But solid image prompting stacks the odds in your favor.
Video generation used to be the same way because you were essentially asking the AI to take a single image and deduce, i.e. guess, what happens next. Fortunately, if you use video generation platforms like Sea Dance 2.0 or 2.5 in CapCut, you can string several images together so that the video prompt can be fairly simple. That is because the images themselves tell the AI video generation platform EXACTLY what the video is supposed to look like. For SeaDance 2.0, you can string together up to 9 nine images to make videos up to 15 seconds in 1080 or 4K video format. In SeaDance 2.5 in 2.5 you can string together up to 30 images for videos lasting up to 30 seconds in 1080 format. The example videos I uploaded were made using 6 images strung together.
I know it may sound like I'm hyping for Google or SeaDance or CapCut like a paid promoter or high net worth corporate shareholder. (I'm not). I just find them to be what works for me right now. If somebody has different methods or platforms or ideas, I would genuinely love to hear them. And we're VERY fortunate to have a forum in which we can share ideas and methods. Again, kudos to Barbarella for setting all this up.
I understand and appreciate that $$$ is a real limiting factor for all of us, including me. This stuff ain't cheap, particularly if you're using the more advanced platforms. But, IMHO, spending the extra money for better tools, ie, advanced software, hardware, etc., saves you money in the long run because you're not generating AI content over and over, hoping for HAPPY accidents.
If we help each other by sharing information, ideas, good prompts, etc.,we will ALL save money for each other.
That's why I'm being so verbose in these posts. I want to help everybody who is interested IMPROVE at this.