16 Comments
Apr 29Liked by Daniel Nest

Hi Daniel! Thanks for this!

I'm having an issue. I created a dedicated server, invited both bots and uploaded two ids to use, but once I'm on the MJ dm and click "More" on the upscaled image the InsightFaceSwap app doesn't show up as an option. I only get the three MidJourney options (Cancel Job, DM Results, Report Job).

Could it be because I'm using MJ v6?

Expand full comment
author

That's odd. I have just tested this with a fresh MJ V6 image that I upscaled, and I get the full InSwapper optioins next to the upscaled image: https://i.imgur.com/7PsSjJg.png

Expand full comment
Apr 30Liked by Daniel Nest

Ok I actually figured out what was happening. It was a dumb mistake but maybe someone's having the same issue? So for me it was that I was trying to use it in the Midjourney dm, where I had already been creating some images, instead of using it in the dedicated server that I created for this purpose (I KNOW RIGHT?).

Anyway, works fine now! Just prompt the Midjourney bot INSIDE the server where the InsightFaceSwap bot is and everything should work as intended.

Thanks Daniel!!

Expand full comment
author

Awesome to hear you figured it out! And yup - that's the main purpose of putting both bots into the same server together. INSwapper doesn't connect to the MJ bot in your private messages.

Expand full comment
Apr 19Liked by Daniel Nest

I'm guessing this no longer works. Once I upscale the Midjourney image, it does not give me the "More" (3 dot) option.

Expand full comment
author

It should still work. Just tested. Are you on a desktop?

Expand full comment

Daniel, are you finding midjourney bots freely available? I don't currently subscribe to MJ, but would love to toy with it a bit on weekends.

Expand full comment
author

They used to offer free trials (up to 25 generations), but I don't believe they do that any longer. So if you want to use Midjourney, you'll have to subscribe. Plans start at $10 a month, but if you're thinking of doing lots of generations, I'd suggest the $30 plan (https://docs.midjourney.com/docs/plans).

Expand full comment

Thanks, Dan! I think I'll end up there before too long. I know where to find some tips, too!

BTW, I really enjoy Matt Wolfe's videos. He's the best YouTuber I've found for showcasing new toys like this.

Expand full comment
author

Agreed, he's clearly passionate about the space and dives into many topics that I'm also interested in! Definitely reach out if you end up using MJ and need any help.

Expand full comment

Wow, thanks for this Daniel. This could turn out to be extremely useful in my current project. I tend to hate the Discord interface for some reason, but your article gives me a good reason to be more open minded about it.

Just yesterday I did this face swap thing the hard way. I wanted a video of Richard Nixon talking in to the camera. It was easy to create a man wearing a suit in Stable Diffusion, but none of the Nixon faces it generated looked like Nixon, and if Nixon wasn't recognizable the video just wouldn't work.

So I grabbed a photo of Nixon and married it to the AI body. It took some fiddling, but I finally got a reasonable result. https://hippytoons.com/p/welcome-to-hippy-toons

But if I could do this operation entirely in AI, it might be far more efficient, and provide many more options. The photos of yourself you generated look pretty excellent, so thanks for sharing those example shots.

Expand full comment
author

Happy to hear you found this useful, Phil! Let me know how it works for you.

If you're looking to animate images, you could try to create the original in Midjourney (+InsightFaceSwap to replace the face) and then run it through Runway's Gen-2 (I demoed it here: https://www.whytryai.com/p/10x-ai-6-gen-2-redpajama-7b-bing-mermaid-code).

Expand full comment

Thanks again Daniel. Best I can tell (correct me as needed) AI video creation is still far behind what can be done the old fashioned way. That will change over time I'm sure.

Also, while I love creating characters in AI, I'm not fully content with static images, I want them to talk and sing, jump around on the screen etc.

I'm getting characters to talk by feeding 'text to speech' output in to a software face animator. If you're aware of AI tools related to this process, I'd love to learn about them. Thanks again.

Awesome article...

Expand full comment
author

Well, there are tools like HeyGen (https://www.heygen.com/) that offer animated avatars that speak your script, but this might be a bit too professional and serious for what you're going for!

And agreed, text-to-video isn't quite perfect yet, but they're making progress fast. I wouldn't be surprised if we can generate realistic videos during next year.

Expand full comment

Am I the only one stuck at photo upload?.. I' adding a picture, typing in my idname, pressing enter and shows an error saying: "image : This option is required. Specify a value", and highlights the file name.

Expand full comment
author
Oct 3, 2023·edited Oct 3, 2023Author

Hi Olga,

This usually happens if you invoke the saveid command and press Enter before you've selected an image. Here's how it looks on my side if I do the same:

https://i.imgur.com/HliQMxN.png

You then need to click into the "Drag and drop" box and select your image. It should then look like this:

https://i.imgur.com/gQuRwbf.png

Now, pressing "Enter" will save the image along with your chosen idname instead of giving you the error.

Hope that works for you!

Expand full comment