Commercially available laser controllers

[fullspeceng]

Also, did GantryCo give you the dithered file or did you just convert this yourself?

Please post the original grayscale image if you have it.

I'm sure I can do a much better dithering pattern than the random one you supplied.

[macona]

We are confident to say our USB RetinaEngrave software and controller is superior to both of these systems. We have all 3 control cards side by side running on the same machine with the same graphics file and can do an apples to apples comparison.

I think you may be speaking a little soon on that statement. Your hardware may be for engraving but the others still have some advantages like a control interface on the machine itself and the ability to change setting on the fly even during a run. Also the chinese controls support jogging, homing, soft limits, and assist air as well.

And until you software can do multiple layers with different power levels and speeds and mixed modes of engraving and vector cutting I can definitely say your software currently does not compete.

And then there is something as simple as configuration. I only know the MPC6515 but it is pretty easy to adjust settings but you need to edit an XML on yours to do any of that.

I know most of this can be solved in software but saying your control is superior to others at this point in time is a bit presumptuous.

[fullspeceng]

We now support color properties in layers and do it superior to others.

See our new videos in HD:
http://www.youtube.com/watch?v=b7UoH-XfYME
http://www.youtube.com/watch?v=DDQHpJlRTic

Basically these are our claims:
1) We have superior vector trajectory to EMC/Mach3 and the Chinese controllers for laser. Our cuts are more even than any of the options.
2) We have superior raster capabilities to EMC/Mach3 and the Chinese controllers for laser. Our raster output is superior quality as shown in images.
3) We have superior ease of use with direct print capabilities. No one else offers direct print drivers. Want to output from MS Word? Tough luck with the other options.
4) We have superior compatibility on Windows 7 x32bit and x64bit.
5) We offer superior general update support. Once you buy the Chinese controllers you will probably never get an update.

We invite you to refute our claims by showing output from your laser and presenting the BMP or Vector file to cut as well as final output. We have already supplied the native BMP file of the Aztec calendar. We ran this on a MPC6515 and it's nowhere close.

We are scientists at heart. Show us what you can do and supply the type of conditions and we will show you how it is done better on our system.

We offered to mail our sample for free to John to compare against his own machine. He can raster on the back of the sample and comment on the output quality.

We even give 100% refunds within 60 days on any control card we sell to current customers of our laser system (note you need to buy our laser for this to apply). We are THAT confident that it is superior to the other options out there.

We conceed the following points:
1) As mentioned our primary goal is to sell to OEMs and our own machines so adjustment for DIY is probably not as good as the other options.
2) It doesn't work out of the box for DIY and will take some time to adjust parameters.
3) We do not have other advanced features as you mentioned yet.

[macona]

Wow, it's really coming along!

Layers looks good. Appears to be almost identical to lasercut but can't see anything that makes it superior. Is there a limit to the amount of layers?

How does the unlock disable the drives? Is there an enable line to the steppers?

Also it appears that the computer can now control laser power?

You say the cuts are more even. What does that mean?

A simple motor tuning interface like mach's would make a world of difference. Even oem's are not going to want to hand edit an XML for every machine. As you say, finding the sweet spot for a machine takes time. Even a rudimentary XML editor would be nice.

Homing and limits are not advanced features. These are basic necessities that all cnc machines need to have to reduce the chances of mechanical damage.

[lasersafe1]

Also, did GantryCo give you the dithered file or did you just convert this yourself?

Please post the original grayscale image if you have it.

I'm sure I can do a much better dithering pattern than the random one you supplied.

The raw undithered 8bit greyscale image is attached. All these images are copyrighted by GantryCo.

Your engravings look good, but I'm not seeing anything that could indicate it they are superior to other engrave controllers. I think you would need to show extreme closeup photos and describe the difference.

[lasersafe1]

I am still practicing, but here is one of my latest outputs of an image I'm making for a friend. (No, I can't upload the raw file, but here is a hint: The border is from GantryCo. The Lincoln face was done using the prep I showed in the video of my earlier post.)
The image was dithered to 1 bit at 900dpi random. I am still testing various dither modes and dpi settings, speeds and powers to determine what is actually required. Sometimes my software will crash if the file is too large. I think my dither densities are too high.
This was done with the LightObjects controller with the following settings:

Power 21watts. (60% of 35W) using a DC tube.
Speed 200mm/sec.
Material: Black Walnut (Note: Should have gone cross grain instead of with the grain )
Stepover: 0.015mm on Y. (apparently this was too large)
Raster: Bidirectional X cut.
Air assist was used (and needed to prevent flameup and sap staining)
Post cleaning was soap and water with toothbrush.

I would say that this output is marketable in its present state, but I am still practicing.

[fullspeceng]

Thanks for giving the grayscale. I will play with it later and show some TRUE 3d results True 3D can be accomplished in a number of different ways. The best way I can think of right now is to use a small stepover and vector cut with varying power. This is the exactly the same way CNC mills will pocket out a taper using a ball end mill. The problem with laser is that the depth is not obviously related to cutting power. Twice as much power is not going to be twice the depth. To figure out this relationship will probably require an interferometer and a LOT of testing. A short focus highly diverging beam should give the best output at any XYZ. Since we have our own controller we can also raise the Z table as a function of the pocket. Some simulations and thinking would need to take place to figure out the best solution. Mach3 should also be able to do pocketing in a similar manner but I have noticed it delays the laser on/off by a few milliseconds and you will probably not be able to get it as clean as your own solution.

Before we started selling lasers/developing laser stuff, we did a number of 3d laser scans and wrote code to make our own 3d laser scanner and also part of our true 3d printer project using inkjet technology. If you have the surface depth map (like from a laser scan) it is very easy to create a true 3d grayscale image.

About the image results:
My point is that to compare 2 lasers you cannot fiddle with those parameters. I asked for the 1 bit BMP image. A laser should always fire if the dot is black and not fire if it is white. If you engrave the grayscale image on an Epilog or Universal laser then the output will probably be better because their grayscale filtering is optimized. However, if you print the dithered 1 bit image that John posted then my next claim is that it will not be any better.

If you engrave that 1-bit image then your Chinese controller should just work. If you run out of memory or whatever then something is wrong and exactly my point about poorly written software that never gets fixed.

I am also not sure why you show the Abe Lincoln image when you posted the gantryco 1 bit bitmap. This is supposed to be apples to apples comparison. As an aside I can clearly see a LOT of vertical striations in your output. I believe this to be a result of too large a step size. I have noticed this on the leetro controller also. To improve the speed of the output, a 900 dpi is down converted to something like 150dpi and then output. My laser is 1000 true dpi which means it takes 1000 left to right passes per inch. There's no reason to downsample the X axis but there are many advantages to downsample the Y axis as the speed goes up linearly with the fewer Y passes.

Shouldn't it be trivial for you to open up the 1bit image, hit raster and get some output? The only adjustable parameter is the power on my setup.

Send me a PM with your address or email it to me and let me send this sample carving to you in the mail. Then engrave the back of this piece of wood with your own controller using the 1 bit bitmap you supplied. My digital photography skills are clearly nowhere close to as good as yours and it is hard to judge the true impact without having both in person side by side. Report back your independent results

Other offers:
I am an expert at image processing for other purposes I cannot disclose and think I can write a much better filter. I am willing to do this work open source and allow everyone to have it if some people are willing to help. Doesn't need to be coding, just let me do the math and you can output it on your laser for testing. Anything so I feel I'm not working alone on this. I will probably write this code in Matlab so it won't be fast or easy to use but just demonstrate that it can work.

It's the beauty of having control over everything from the hardware to the electronics to the software. Anything you want to do you can without relying on someone else.

[lasersafe1]

There are horizontal striations, not vertical, and they are primarily due to the grain pattern of the walnut, but yes, the laser will also add to this. I agree that any improvements to a conversion algorithm would be interesting and worthwhile.

Yes, I am looking forward to seeing your outputs in high detail. ( I use a simple Sony Cybershot handheld 7.2MP camera turned down to 2MP. It has a good MACRO focus function)

If you look back in the thread you will see that engraving methods were discussed quite a bit. For the most part I came away with an impression that it simply cant be done well with a DC tube and power supply. The best units are using an rf laser and changing the power based on color depth in real time. Still, I'm starting to get greater satisfaction with my DC tube and controller now that I am playing with the settings. I once thought that the burn would create a sponge-like surface when it is simply on-off, but my results are showing a rather solid structure of varying depth.

I thought you had the ability to show us apples to apples. I thought you were going to run your file on a machine with your controller and another with the Chinese controller. You mentioned that the dither was a problem the way I did it because the controller would overstrike the same pixel twice. That is not true if the overstriking mode in the controller is set to "unsupress". I haven't played with the other settings, so I don't know exactly what they do, but the overstriking mode can also be set to Intaglio or Rilievi mode. (anybody understand those modes?)

I will download and scan the mayan calendar from your site.

[lasersafe1]

Here are a few more test results and their associated files:

Forgive me for not maintaining a constant size format between my image and the engraved output. I generally like to keep my images larger so they keep the resolution, but then I shrink the output right before engraving because smaller engravings usually look better in wood.

These engravings were produced from lincoln1.jpg with two different raster speed settings and 40% power.

Now Don't laugh. I had to try this....Notice that in the first images above, the engavings looked pure, but they only had a visual depth contour and no "real" contour as you would find in a real statue. In order to correct this and add contour, I hand processed the image to bring the things that should be closer forward with white and sent the things that should be further back to the rear with shading. The result now has a rounded surface, but it is a little "clown like". This was done with negative lincoln2.jpg

Indeed, even the image is clown like, but I thought that it needed some bold variations to show depth better. Perhaps I went too bold.

This engraving gave me a little more of what I was looking for. It has real contour. The nose, chin, hair and tie are closest, and the face curves away as it fades to the rear.

This was not easy, but with practice I think I can get better. You have to learn to think negative as you add or remove intensity. My three favorite tools in Photoshop are dodge, burn, and Sponge. The sponge is very interesting because it can move the color toward or away from saturation.

Notice that the close items are a little too close because the raw wood appears. This is easy to correct in the future by taking the whole image down a couple bits away from white.

here is the file that did the engraving.

These latest files are mine completly and I am releasing them to the world without copyright.

[lasersafe1]

I finally engraved the Mayan (or is it Aztec?) who cares because they're all dead now.. calendar in high detail.
I didn't even try to clean this thing because the parts are so small I'm sure they would break.
First I will attach the test file that I used directly with no scaling. Both the graphic and the output are 2.5 inches.
The resolution of the graphic is 600dpi.

Mayan (Aztec) test graphic

Mayan burn output

Notice if you click on the picture above and zoom in on the two warriors on the bottom, you can even see their teeth.

I think the wood is Maple, but I'm not sure. It has a very tight grain. The burn depth is not evident in the photos, but all black spots go down 1/8 inch.
Raster settings were 200mm/sec. 25% power (of 35W). 0.005mm interval step on Y.
I used unipolar sweep mode for this run because I feared any belt slop would cut away the small pillars.
At this feed and speed, it took 2hours 10 minutes. This is not something I would do very often. Perhaps I will try another with bi-directional sweep, larger Y step interval and faster mm/sec. At least I have this documented now.