• It's time to cast your vote in the March 2025 Turning Challenge. (click here for details)
  • Congratulations to Steve Bonny for "Rhonda and Fisherman"being selected as Turning of the Week for March 31st, 202 (click here for details)
  • Welcome new registering member. Your username must be your real First and Last name (for example: John Doe). "Screen names" and "handles" are not allowed and your registration will be deleted if you don't use your real name. Also, do not use all caps nor all lower case.

Signatures, Laser Engravers and Generative AI

Joined
Dec 28, 2019
Messages
634
Likes
814
Location
Tallahassee, FL
This is more of a tip than a tutorial—something along the lines of "don’t do what I did. ” :) Like many others here on the forum, I have a laser engraver. It’s a small 5-watt model, but since I’m not very artistic, I’ve been using one of the generative AI models—in this case, Copilot—to create artwork for my signatures to use on my bowls and boxes.
Several generative AI models, like ChatGPT or Copilot, can create artwork based on your directions or prompts. For my signature design, I wanted a simple logo with space to insert my signature and an alligator to represent Florida. Below you can see one of the prompts I used:

"Draw a black and white circular logo with 2025 at the bottom edge. Include an alligator silhouette and a lathe in the logo. Leave space in the logo for my signature. Put the logo on a empty white background with a clean round edge. "

Unlike using a search engine, you’ll want to be as descriptive as possible when working with generative AI. However, some tools, like Copilot, have a character limit for prompts. Despite the limited number of characters, my directions are fairly detailed. I include instructions on the shape of the logo, the colors to use, and the design elements I want included. This is a sample of what I got:

CoPilot Screenshot.jpg


Visual generative AI is like a slot machine—you never know what you’ll get, and I haven’t found a way to correct an image. You can tell the AI to change something specific, but it doesn’t really do it; it just re-generates the image, which may or may not address the issue with the original. The only option is to re-run your prompt and take your chances on what you get. The advantage of generative AI is that it’s incredibly fast and cost-effective. You can generate dozens of outputs in a relatively short time frame, and if you’re artistically challenged like me, you can achieve a quality of work that you would otherwise need to commission from someone else.

To get a few logos I was happy with, I had to re-run my prompt several times. I rarely receive something I’m 100% satisfied with, and while that may sound discouraging, I usually get something I can use fairly quickly. It might require me to open MS Paint or PhotosShop for a simple edit, but for me, the time (and cost) savings are worth it. Think of the output as a rough draft—though that idea may become outdated in a few years. It’s something you can use but might need to polish before it’s ready.

Also, visual generative AI is bad at spelling. Even if you include the correct spelling of a word in your prompt, the AI often won’t spell it correctly. This is one of the reasons you might need to re-run your prompt. Additionally, it may add words from your prompt directly to your images. You can fix this using Photoshop or by re-running your prompt.

CoPilot Bad spelling example 1.png CoPilot Bad logo example 1.png

So after about a dozen tries, I received a few images I could use like the one below.

CoPilot Bad logo example 2.png

It still wasn't perfect but after using MS Paint to touch up the image I had something I could use. Since I reached my photo limit I'll post the finished images below.

Generative AI is great for creating artwork but like all tools it has limitations. It's advantages is that it's quick and can generative many images in a few minutes but you may still need to edit them when you're done. As the tools get better creating logos and other images to use for woodworking will get easier to use.
 
Something I forgot to include, once you have a prompt, for whatever generative AI you're using, save it. Either write it down, print it out or safe in it a file to use again. You can use it as a template the next time you're trying to create something similar. It will save you time and the AI should create something similar to what it did last time. Some of the tools have the ability to save conversations or threads so you can use this to create similar output to what you did previously but I would still save the prompts you created somewhere else in case you lose access to that tool.
 
Thanks for the thread. Anyone else been playing around with this? I have some AI software that generates .svg files that can be easily edited or resized. Like you Charles, I find that you have to be quite specific with the prompts and results are very sporadic. It is fun to see the crazy things the AI comes up with and some of them are even useful. Sometimes I try prompts just to produce something as inspiration for a piece. For instance, the prompt "Japanese style hollow form with a carved tree" produced the attached image. This exact design may not be a viable project, but it might be inspiration for something I do actually attempt one day.

JStyleHollow1.png
 
Good idea! If you get an image you like, you can bring it into a free program like inkscape and trace it. From there you can update text for the year and clean up any AI hallucinations.

I got a laserpecker LP1 last year and have been using a simple circular text design I did in inkscape. Maybe I will switch it up for 2025 and see what AI can suggest.
 
@Charles Cadenhead

Thanks for the thread. I have a Laserpecker 2 coming and one of the primary things I will be doing is using it for my signature. Using the brand is ok but I can never get it uniform.

Bob
I've had a lot of fun with mine. I'm really impressed with what it can do, though I wish they would make the app more user friendly. If you haven't seen them, there are several posts here from turners who mounted the laser on their lathes for burning.
 
Thanks for the thread. Anyone else been playing around with this? I have some AI software that generates .svg files that can be easily edited or resized. Like you Charles, I find that you have to be quite specific with the prompts and results are very sporadic. It is fun to see the crazy things the AI comes up with and some of them are even useful. Sometimes I try prompts just to produce something as inspiration for a piece. For instance, the prompt "Japanese style hollow form with a carved tree" produced the attached image. This exact design may not be a viable project, but it might be inspiration for something I do actually attempt one day.

View attachment 70363
Darryn, I have seen a lot of AI-generated images similar to this and other woodturning projects online, especially over the past three or four months although yours is probably the most intricate I have seen. I would like to play around in this area but I would need some basic instructions or to be able to watch over someone's shoulder for a bit before striking out on my own. I haven't been able to find the help I've been looking for.

Charles, thank you for the examples you posted on making a logo.
 
Darryn, I have seen a lot of AI-generated images similar to this and other woodturning projects online, especially over the past three or four months although yours is probably the most intricate I have seen. I would like to play around in this area but I would need some basic instructions or to be able to watch over someone's shoulder for a bit before striking out on my own. I haven't been able to find the help I've been looking for.

Charles, thank you for the examples you posted on making a logo.
I see a lot of AI images online that people try to pass off as actual turnings. Most of them are obviously AI, but as the technology improves it will be more difficult to spot them. AI is a fun and useful tool but as with any technology some ethics should be involved. I find the use of AI fairly intuitive due to my background. I have been turning logic and computer language into things visible on a screen for more years than I care to acknowledge.
 
I wonder if that image was a picture on the internet that AI found (dare I say plagiarized), rather than created?
It was generated. Generative AI doesn't use whole or even parts of works. For pictures and videos it works on the pixel level for text it can use words, syllables and characters.

One of the ways you can tell it was generated is by visual anomalies, like too many fingers, smooth or rounded edges and inconsistent lighting and shadows. If you look at generated pictures enough you start to notice them.
 
I have used many AI generated images for some of my other crafts. This past year I made laser engraved trivets with 6" white tiles (used the Norton White Tile method). I must have close to 100 images of various subject matters. (I use the Microsoft image creator). I had some people question me on the images and how long it took me to draw them. When I told them how I generated them some were curious, others were put off and said if they were not "hand drawn" they were not interested. I am fine with that thinking, but I would not be selling hand drawn artwork for $8, especially since typically I would have 50+ trivets at the show and maybe only 2 or 3 duplicates.

Most AI image generators have trouble with text and numbers. I use Inkscape (or Lightburn) to add in any text or numbers I need after the image is generated.
 
Most AI image generators have trouble with text and numbers. I use Inkscape (or Lightburn) to add in any text or numbers I need after the image is generated.
Thats what I ended up doing after trying a dozen times to get a logo as like Charles did, I finally gave up on that and just had it generate a coin-sized logo with no text, picked something with decent embellishments where my text could fit and then edited the result in GIMP to add my desired text.
 
I've been using Microsoft/Bing Image Creator to generate AI art for about a year, then mod and clean up the image using Lightburn. Here's an example of the AI output from these prompts: "Chicago skyline, beautiful sunset in background, black and white, detailed line drawing, no gradients, centered, well-balanced, negative space"
Screenshot 2025-01-03 085955.jpg

And here's the finished product. Maker's mark in lower righthand corner added with Lightburn, along with other clean-ups and tracing for 10w diode lasering:
2024_yoyo_174e.jpg
 
Last edited:
I've never played around with AI image generation but I'm curious. Is the Bing service a good place to start, or are there other free options I should look at?
 
I've never played around with AI image generation but I'm curious. Is the Bing service a good place to start, or are there other free options I should look at?
All of them are about the same, but Microsoft's image generator is pretty good. You can find it at https://designer.microsoft.com/ Under the "Create with AI" menu you can pick what types of images you want to generate.
 
I have a Laserpecker 1. I use it to”burn” my logo onto my work. Not AI generated, made it using powerpoint and saving it to an image.
JD-Logo - square v8 cropped.png
IMG_8918.jpeg
 
I have a Laserpecker 1. I use it to”burn” my logo onto my work. Not AI generated, made it using powerpoint and saving it to an image.
View attachment 73115
View attachment 73116

Couple of thoughts if you don't mind. I'd get rid of the shading and grain on the tools. They come out unrecognizable. I would consider getting rid of the bounding box. It clashes with the detail rings.
 
Thanks for the recommendation.
So you have separate eye protection? (since it looks like it doesn't come with any kind of shielding). Also, do you need any fans or anything to deal with smoke?
 
Thanks for the recommendation.
So you have separate eye protection? (since it looks like it doesn't come with any kind of shielding). Also, do you need any fans or anything to deal with smoke?
I think they came with a pair of safety glasses (or maybe Im thinking of my cheapie 80mm laser) , since I largely use mine lathe mounted (Just made a base to bolt it to that screws to a 1" dowel that goes in banjo like a tool rest - 1/4-20 thread, BTW) I don't bother with fan since dust collector nozzle can move air and suck smoke away - also smoke has little effect when I'm using it mounted in lathe to burn logos into bottom of bowls, etc since it rises upwards above the working laser beam...
 
That is a great idea. I also want to engrave the bottom of bowls I turn, and I had not thought about mounting the laser on the lathe banjo. That eliminates any smoke issues, and holds everything steady.
I don't have a vacuum chuck, so I'd have to think of another way to hold the completed bowl (may be tape?) but that should be a solvable problem.
 
Thanks for the recommendation.
So you have separate eye protection? (since it looks like it doesn't come with any kind of shielding). Also, do you need any fans or anything to deal with smoke?
It came with a pair of glasses. The smoke is so minimal, I haven’t worried about it. I thought about mounting it on the lathe but find it easier to just set the bowl on the bench, setup the laser on its tripod, use the preview function to get the position where I want it, and let it burn away. Just have to remember that it engraves upside-down.
 
That is a great idea. I also want to engrave the bottom of bowls I turn, and I had not thought about mounting the laser on the lathe banjo. That eliminates any smoke issues, and holds everything steady.
I don't have a vacuum chuck, so I'd have to think of another way to hold the completed bowl (may be tape?) but that should be a solvable problem.
Other benefit to mounting to banjo is you can more easily adjust for centering and distance (200mm from laser head to work surface) - as far as holding the bowl in place, it'd be fairly simple even as simple as making a sort of jam chuck and a bit of double sided tape to stick bowl to , since you're not spinning up the lathe or anything and all that is needed is to hold it in place for a minute or two.. Me, I have used cole jaws as well as jam chucking with double sided tape for bowls, or just reverse chucking for boxes, etc - almost unlimited options can work - whatever gets the job done...
 
I have a Laserpecker (1?) Pro for signing pieces. I like that I can also do logos, indexed repeated burns on rims, and inscriptions or art on the insides of bowls. It's not real powerful, but it does a decent job. It says the maximum size is about 4" square, but that can be stretched a bit with a longer distance away. (with some loss of resolution)
 

Attachments

  • PXL_20250220_204854551.MP-EDIT.jpg
    PXL_20250220_204854551.MP-EDIT.jpg
    632.4 KB · Views: 18
  • 315960566_213768647742500_1475034135211558202_n.jpg
    315960566_213768647742500_1475034135211558202_n.jpg
    144 KB · Views: 21
  • IMG_20230201_211948431.jpg
    IMG_20230201_211948431.jpg
    400.7 KB · Views: 19
  • IMG_20220714_171600739.jpg
    IMG_20220714_171600739.jpg
    416.5 KB · Views: 21
I have a Laserpecker (1?) Pro for signing pieces. I like that I can also do logos, indexed repeated burns on rims, and inscriptions or art on the insides of bowls. It's not real powerful, but it does a decent job. It says the maximum size is about 4" square, but that can be stretched a bit with a longer distance away. (with some loss of resolution)
Beautiful work!
 
Back
Top