Some years, my wife and daughter make gift tags. Typically, they are made using rubber stamps and hand coloring the result, but they have done other things. I decided that, this year, I’d try and make some tags using my tools rather than have inkjet-printed-at-the-last-minute tags.
I ended up spending far more time that I should and creating far more designs that I needed. I received a lot of joy making these tags. The thin wood (1/16") cut substantially faster and held detail better than my typical projects. The smaller size meant that engraving was faster (than my 4" tiles). And the simpler designs for the small format meant that graphics work was quicker and far more forgiving. My original thought was to make traditional to/from labels, but I found I had more “canvas” to work with if I put the message on the back.
All of these tags were created on my Primo. Almost all the graphics came from images I “created” using Midjourney.
Those are incredible, Robert! You amaze me with your use of Midjourney to come up with fantastic graphics. I piddled with it a bit, way back when, but was never able to get the suitable-for-laser graphics that you show. I’d love to be able to do some of these myself. Have you created a “collection” of them that you’re willing to share? Those would make a great ArtLibrary collection in Lightburn.
I’d be glad to provide a set of Midjourney images for you to use, though I’m out of town, so it will be a few days before I can post the graphics. Are you looking for something specific? I tend to do Midjourney prompts for animals, flowers, trees/forest, holidays, and fandoms (superheroes for example). Note I usually used Midjourney prompts that tend to produce images that either are, or can be made to be, two-toned black and white. My laser engravings are done using threshold rather than any dithering. FYI, for engraving, I treated the wood with borax to darken the burn color.
As for Midjourney, I’ll be glad to provide you with some prompt examples that, for laser use, have produced good results for me. All the images used for the tags above were generated from very similar prompts with only minor variations.
For engraving, Midjourney images can usually be used with little or no manipulation. For laser cutting, there is almost always Photoshop work to be done, and sometimes vector work in Lightburn as well.
Those look fantastic. I never thought about using one of the LLMs for this. I’d be interested in some prompt examples and a little more detail on the process if and when you have time. This might speed up my flow a lot since I don’t have a lot of time to spend creating the graphics (though I do enjoy that part of the process). Thanks for sharing.
Nothing specific, Robert. I was just blown away by the quantity, quality, and variety of gift tags you showed. I’ve not looked at Midjourney lately but IIRC I had only a limited number of image attempts – and without any of the successes you’ve shown – before they demanded I start paying for the further use of their product. And I wasn’t willing to commit $$$ without anything useful to show for my initial efforts.
I’d, of course, love to have a set of the Midjourney images you’ve done. [Have I mentioned I think your work is nothing short of phenomenal?] I, too, have a crafty daughter and, as I said, the set you’ve shown would make a fantastic Art Library for Lightburn (and I suspect that people would be willing to pay for it, packaged that way). But, of course, my initial unsuccessful attempts at using Midjourney for laser-suitable images was without doubt my not using suitable prompts… so anything you could share on that front would be deeply appreciated as well.
The red and green backings, and even the black backings, are obtained by gluing posterboard of those colors to back of the tag. I first tried a second layer of wood painted in the color I wanted, but it made the tags too thick (IMO). On the sixth image above, the black backing behind the deer is wood that has been darkened using leather stain, and the Santa with red had is white poster board with red sharpie in the area of the hat.
All the more complex graphics came out of Midjourney, though as I mentioned, there is usually some post processing work to get the graphics to a point I can use it for lser cutting. There seems some interest in Midjourney to generate graphics, so I’ll post a separate topic on the subject next week (when I return home).
You spray a solution on the wood. If you search YouTube for borax laser engraving, you will find a number of videos. I purchased a box of 20 Mule Team Borax from the laundry department at Walmart. I heated 5oz of water in the microwave (hot to the touch but not boiling), and added 1 Tablesoon of borax. After cooling I put it in a mister bottle and sprayed the wood.
According to tests on YouTube, borax only works on some woods…typically woods of lighter color. The plywood I used for the tags was basswood, and borax works particularly on this wood.
I did the tag engraving over a period of days. I noticed that the borax crystalized out of the wood, and the later tags were a bit uneven. It is probably best to treat the wood and then use it within the day.
Note the dark color is soot on the wood. It can easily be smeared, so it is best to handle the wood carefully and then clear coat the wood to lock in the color.
Fine details and isolated islands are two of the main issues to be worked around with Midjourney. I’ll created a separate topic in the next few days with some of the prompts I’ve found useful.
I wanted to make a small present for my daughter’s nanny, and she just happen to have installed the christmas tree this week, so I made this little ornament with the names of my daughter and the other kid she’s taking care of
Image generated by Bing (Dall-e)
I think it took something like 20 prompts to get a usable image
“nanny nith 3 year old boy and girl playing, black and white single line inside a circle”
I’d love to be able to do some of these myself. Have you created a “collection” of them that you’re willing to share?
I’ve been struggling with how to share some Midjourney files. Here is a dropbox link. I’ve never used Dropbox links, so I don’t know if this will work. Please let me know.
I selected files that related to Christmas or winter. There are nearly 1,500 files in the share. This is just a portion of the Midjourney images I produced over about a 10-day period. I was focusing on Christmas and winter during this period.
There is a mixture of cutting and engraving images here. There are lots of different takes on Christmas/evergreen trees. Even when I’m trying to produce images for engraving, I tend to use prompts that force the image into black and white with little gray. For laser cutting, I expect most images will need both bitmap editing and vector editing (in Lightburn). I use an old version of Photoshop for bitmap editing or I will use the free, online Photoshop clone Photopea. I would guess I spend, on average, 45 minutes going from a “raw” Midjourney image to a Lightburn vector version. I extract the image into a two-tone, BW mask, remove elements, bridge or otherwise deal with islands, and bulk up narrow areas.
I’ve been playing with using an online backround remover to speed up the process. I’ve settled on this one. It seems to do a better job than the other removers, and it downloads the results (for free) at a higher resolution…and it does not require an account. Results vary, but usually it speeds things up.
I can’t/won’t leave the link active forever, so grab whatever you want sooner rather than later.
Gimp (Gnu Image Manipulation Something) is the big open source project that competes with photoshop (the way inkscape competes with illustrator). I’ve used it for years and I am a little dangerous in it. But I haven’t used actual photoshop since college.
Goodness, Robert. I didn’t realize the extent of your collection and the ends to which you had to go to share it. I thought I was asking for just a sampling of the examples you shared in the first post and a few of the prompts required to get similar results. I deeply appreciate your efforts and was able to take a look at your collection and copy a few that jumped out at me. Again, I deeply appreciate your generosity and willingness to share your work. Thank you!
My problem seems to be that the prompts/keywords I’ve used have resulted in far more image complexity than I’m looking for. Your images all seem elegant without being overly so… and are thus striking and unique as gift tags and ornaments. I suppose I just can’t get my head wrapped around using these incredibly powerful AI tools to generate what is essentially glorified clipart… compared to the wildly complex and realistic images they are actually capable of.
Every few months (when I know I’ll have daily time to invest), I pay for the $30 standard Midjourney plan. At this plan level, I can do as many “/relaxed” level images as I want, and it is not unusual to do 200 to 400 generations per day (each generation produces four images). Note this is not constant Midjourney work. I just load up the queue with prompts and then regularly visit my computer between other tasks to add another set of prompts.
I find using Midjourney more like mining than artistic creation. I start with a promising prompt, and then mutate that prompt to see what I can get. If I find something that I like, I’ll often run that exact prompt 20 or 30 times. I’ll then go back and upscale anything that is promising as reference art, or if there is some interesting concept in the panel. While it varies with prompt and the subject, I upscale about 1 in 4 of the images created, and maybe 1 in 10 to 1 in 20 of those images are easily converted for laser cutting. That works out to around 1 or 2 images per hundred…which doesn’t sound like a lot, but it is easy in Midjourney to generate 100 images to sample from.
I paid for a month of Midjourney standard plan from October 8th to November 8th. In that time, I created collections of:
Halloween
Animals
Flowers
in the style of Kliban
Christmas
Winter
What I shared via Dropbox was a selection (maybe 1 in 10) of the images where “Christmas” or “winter” was in the prompt.
My problem seems to be that the prompts/keywords I’ve used have resulted in far more image complexity than I’m looking for.
You are exactly right. The trick is to try and control the complexity. Fabian above asked for prompts. I haven’t forgotten and will make a post or new topic with some of the prompts I use. It is not as simple as just making a list of prompts.
AI tools to generate what is essentially glorified clipart …
Unfortunately, as mentioned above, going from the Midjourney image to vector clipart takes editing time, so making a Lightburn library of them would be a major time investment. Personally, I use the art to make gifts, so it is rare for me to use a vector image more than once.
Concerning the prompts, you’re totally right
I don’t expect any “pre-made” and ready to use prompts, just tips about keywords that work for this kind of image
The whole process you described above is definitely very valuable, thanks!
Also, I’ve seen some people recommanding using chatgpt to generate prompts for midjourney, I might have to try that…
