I’ve been working on figuring out a recipe for laser engraving white tiles. I’m still tossing more tiles than I’m keeping, but I wanted to share my keepers so far. But first I want to thank David (@dkj4linux). His repeated, detailed instructions, presentation of quality results, and a pointer to the JL1 were all extremely helpful.
Some of these tiles were engraved using my Primo. Lately I’ve been using the JL1 since it is somewhat faster. The results are similar, though not identical, between the two machines (same Lightburn file, same laser). All used a NEJE A40640 laser. All but one tile was engraved at 300 LPI.
All the images used for engraving were created by the Midjourney AI image generator.
Those are fantastic, Robert! And I was wondering where you were getting those neat little animal pictures… both weird and mesmerizing at the same time. Midjourney AI image generator…never heard of it (not surprising!) but need to look into it. Wild stuff!
My art/organic creation ability is extremely limited. My use of drawing and sculpting tools (real or computer), only produces poor, primitive works that often don’t look right, but no matter how much I thrash around, I cannot fix them. On the other hand, I do pretty well recreating and/or modifying some other work I can see.
I’ve not been using Midjourney to produce art like you see in the video above. I’ve been asking it to produce things that I can 1) laser cut, 2) laser engrave, and 3) relief carve. In the case of laser cutting and relief carving, there is a significant amount of work to transform the images Midjourney produces in to vectors files or 3D models, but that work is pretty straight forward.
I think that is a really clever solution, Robert. I haven’t seen anyone trying to make vector graphics or design files from these tools.
The tech can obviously do a lot more. But it is great to see it used in the more “simple” case.
What do the images coming from MJ look like? Are they pretty much black and white or does the bitmap to svg process do most of the conversion?
I could easily imagine a lot of the etsy/cricut/craft faire designs being built this way. Prompts like “Live, Love, Wine, cursive in the shape of a wine bottle” would work pretty well after some iterations and commands to encourage the binary image style.
It is amazing and scary at the same time. I’m shaky to the point of hardly holding a pencil and not artistic at all. A couple of minutes and this description “cute frog, portrait, b&w clipart look, rough skin, show front feet” gave me this little guy (and variations) that I can actually laser onto a tile. Not the greatest but amazing still. Lasering the tile as I write this…
If you use /settings, a list of default settings will appear. There are different versions of MJ engine. The latest two (alpha or beta versions) are “MJ Test” and “MJ Test Photo.” Both produce substantially richer and more detailed results. For example, if I type in your prompt, I get this:
Also, below your enlarged image, there may be a “Remaster” button. I think it takes MJ Version 3 images and reimagines them as MJ Test images. It usually produces a richer result.
Last post… don’t want to hijack your thread. Thank you, Robert!
Obviously don’t know what I’m doing but following your instructions as best I could… wound up with these characters. Used "MJ Test " and my exact prompt from before…
No clue where the jacket came from… but he’s still cute
This video describes many of the settings/features/modifiers used by Midjourney. It does not cover some of the recent changes like the MJ Test setting.
This technology is awesome for the pure of heart but I can see it being used by some not so pure characters in a lot of nefarious ways. Most people already don’t do any fact checking on stuff they find online and this will make it nearly impossible to tell the difference between a real photo/video and a fake.
I also don’t know who would actually be the content owner. A person enters a few words/phrases and the AI generates a masterpiece. Does the person, the AI, or the software developer own it or is it now a new form of “Open Source Art” that’s not really owned by anyone?
It’ll be interesting to see how/if these issues will be addressed as the technology matures. I’m excited to see how it will be adapted to other industries such as architecture, automotive, or aviation. Combine it with other AI driven CAE tools and you could potentially cut the design cycle from decades or years to months or even weeks. This could drop cost and accelerate development. But as we learned with the advent of the compact disks (and numerous other examples) the corporations are typically too greedy to pass the savings onto the employees or consumers.
But overall I think this could be the start of our next technological leap so hopefully it’s used more for good than evil.
Midjourney does place both content and people limits on their engine. For exampIe, I don’t believe it will image most politicians, and I’ve had some of my prompts kicked back. I started a prompt with “A CNC cutting…,” and it would not allow the word “cutting.” And there are limits on sex and nudity. But, these are all voluntary limits that the company could change tomorrow, or another company could pop up without these limits.
I also don’t know who would actually be the content owner.
Midjourney gives the user liberal rights to the image, but they retain the copyright to the image. So, if their licensing agreement holds up, they are the owner of the images. There are ways that a bigger company can spend $$ to get the rights.
Another aspect to the engine is that, unless you pay extra to the tune of $20 per month, all your work is public. That doesn’t mean it would be easy to find. This public aspect also applies to the prompts used to create the art. Other can just borrow your prompt, though given the randomness, it may or may not produce a similar result.
I think the $50/month tier is where you get to keep the copyright. Midjourney doesn’t really own it, but when you use their free and cheaper tiers, you are agreeing that they get to have the copyright.
There is a larger question about if either of you should get it if the computer was trained on copyrighted images from the Internet. But that is all very murky and hasn’t been tested or decided.
Honestly. It doesn’t do anything photoshop can’t do. It just does it in 30s instead of three weeks. We will all have to get used to the fact that images are going to look real, and be completely fake. But we are already used to that, right? It won’t be long until someone builds this into a darkish web app that does the same thing without the restrictions. MJ and Dall-e are setting the standards and self regulating. But that won’t be permanent. That is the cost of progress.
It will definitely hurt the struggling artist making ends meat by making art for gizmodo. Just like Internet shopping hurt the salesperson.
AI has been around long enough that there are many, especially those who work with it closely, who have seen fit to warn us of the dangers inherent in its abuse. Elon Musk is one of those prominent voices. AI is here to stay, however, and I’m sure we’ll see examples of both its USE and its ABUSE. Personally, I think that it’s never the THING… but in the ABUSE of the THING where the problem lies. Just add AI to the long list of things that can be used and abused – guns, knives, food, alcohol, drugs, money, power, hammers, airplanes, cars, highways, and anything else you can name… and it is always in the ABUSE of the thing where the problems lie.
Obviously I’m excited about “discovering” this tool and technology. I love the simple but mind-blowing images that Robert has shown that are absolutely unique and “caused” by his use of a few simple words/phrases. I’ve spent hours surfing to find striking images that I can use without fear of abusing someone’s hard work… and now I finally have a tool that allows a shaky non-artist like me to offer “works” guaranteed to be both fun and original. I promise not to abuse it.
