Based on Mud Gecko's Muddy Model Adventure

[lewt_00]

pic (qw) pm mud gecko muddy model (mpb4) 004.jpg

My great respect to Mud Gecko's photomanip The Muddy Model.

I want to present my AI version on this masterpiece.

[Viridian]

I do remember this photomanip. I actually had been planning on doing a full AI remake myself at some point. It's an unbelievable "what if" to bring these to life.

That said, I personally hold a higher standard for transforming these old images. Anyone can upload the image into an AI animation app and let it do its thing, so I prefer to pay homage to the creator by doing more to enhance the original creation. Here's my take.

[BogDog]

Love it.

[lewt_00]

I do remember this photomanip. I actually had been planning on doing a full AI remake myself at some point. It's an unbelievable "what if" to bring these to life.

That said, I personally hold a higher standard for transforming these old images. Anyone can upload the image into an AI animation app and let it do its thing, so I prefer to pay homage to the creator by doing more to enhance the original creation. Here's my take.

Good recreation. However, you said 'anyone', but the AI generators have their quirks. Additionally, I didn't want to embellish the model. I wanted to utilize the the available material to make a captivating vid.

[BogDog]

True that just offering a photo and letting the software auto-prompt it and then render off that generated prompt will almost every time result in unacceptable renders. Waste of time and a waste of money. You have to tell AI what you want it to do and in the right way too. An image is only a part of the instruction. Your text must complete it and in a manner the software can understand.

[lewt_00]

True that just offering a photo and letting the software auto-prompt it and then render off that generated prompt will almost every time result in unacceptable renders. Waste of time and a waste of money. You have to tell AI what you want it to do and in the right way too. An image is only a part of the instruction. Your text must complete it and in a manner the software can understand.

Fair enough.

I did enjoy your version as a dry sinking. Ever since I saw that
Lost in Space episode and Batgirl into the caviar, the dry sinking
works for me.

The two pics emplaced in Kling maintained the model as I wanted,
and I got lucky with the semantics too.

I have not always been successful with Kling especially
when my subject tends to sit. However, my version of
the Muddy Model is almost an exception.

[BogDog]

I hear that. I stopped using the word SINK because the model either squatted down or sometimes would sink that was practically a free-fall. Now I use words like EMBED and PENETRATE to achieve some sink scenes as desired. phrases like "sinks to her knees and stops" don't result in that happening. They just keep going down.

Frustrating some times but also fascinating and admittedly addictive as you wait to see what is rendered per your input. Almost like gambling

[JSample]

Yes, it is addictive and definitely feels like gambling, even just on static image generation.

[lewt_00]

I hear that. I stopped using the word SINK because the model either squatted down or sometimes would sink that was practically a free-fall. Now I use words like EMBED and PENETRATE to achieve some sink scenes as desired. phrases like "sinks to her knees and stops" don't result in that happening. They just keep going down.

Frustrating some times but also fascinating and admittedly addictive as you wait to see what is rendered per your input. Almost like gambling

Thank you for this tip. I will give it a try in future vids.