Slacking today to give my eyes and swollen ankles a chance to recover. Too much time in front of the screen has consequences.
By now most of you know I am using Facebook as a Dropbox for images I create on my phone. It’s that or mail them to myself which I also do. It’s the old school way to move files on the internet without using FTP (the old modern way). It works and it’s cheap and I am a fan of cheap.
Since last December I’ve been absorbing a massive upgrade in hardware and software technology with a brief uneventful trip to the hospital (I’m fibbing. It was terrible but ).
Getting the sound sample library up was straight forward. The learning curve was mostly exploring sounds and experimenting with layers of audio files and direct use of midi as a track in the DAW.
I still hesitate to edit midi in the DAW grid. Old habits die hard but it turns out Guitar Pro 8 has a decent sound engine so I get midi, a notated score and audio files I can layer into the DAW from my midi/tab editor. Cool. In Psychedelic Duck, the drums are two different kits and three different basses (two midi and one audio). Kicks ass. Not as wimpy.
The AI animation isn’t difficult but there are lots of tricks to creating semi consistent images for continuity and figuring out how to get the video generator not to hallucinate. Don’t over specify and avoid excessive labeling when using the image to generate the video. Using a text to video is always dicey but the software can surprise you.
Do remember the physics engines differ a lot among products so be careful about asking a character to exit a room. It may walk through the door. This is when knowing how to use your video editor trim functions is fabulous because you can save part of the clip and points in your generation budget. You have to learn that anyway because the generated clips are specific lengths and when assembling sequences in your video editor you may want to change those.
The software guardrails can be tricky because it isn’t what you think. It refused to generate an image of a Saturn 1 doing a static firing because that is “sensitive information”. So a sixty five year old technology used as a Christmas tree pole in Huntsville is “sensitive“ In a text prompt? No matter. Find an old NASA photo and generate from that. Problem solved. However if it sees a nipple, it says oh hell no. There are AI apps for that but I don’t use them. Dealers Choice.
On top of that I needed and still need to master the Pinnacle video editor. It has a lot of features but a a terrible GUI so finding features is a chore.
I pleasantly found features in the Windows 11 Paint and Photo editors that enable me to process and composite images. All hail to background removal and green screen as well as the workaday color and sharpening features. Compositing images is a must.
Green screen background removal in Vivago.ai lets me extract animation for reuse such as the alien dancing with Godzilla in Psychedelic Duck. BTW the video I posted of me gnoshing with Santa is my way of showing you how easy and fast one can create fakes. Take this seriously, friends. That took one photo and ten minutes. Word.
Anyone who tells you you can just write a prompt and get a perfect work hasn’t tried it. Easy things are easy and some hard things are a lot harder. In any case doing small projects and learning a little from each one is working. Hello world.
Also, if you use AI generated graphics in your art, label them as such. It’s polite to let people know what they are consuming.
I can see a time coming when local studios will need to work with video and film in their sound production. It’s coming.
Time to eat dinner.
