Hi David; thank you for your response. I am sorry if I lost you. I will try to deal with one single issue at a time and that should help the clarity of this debate. You have pointed me to a linked presentation of a cat being etched by laser. I am going to paraphrase your words so that you know how I have understood what you have written. If I am mistaken then I apologise and we should start again.
You say this is the laser operating in its grayscale mode and it is doing exactly what the halftone printing process is trying to do. If that was the case, I would expect the greyscale images to be produced with a type of accuracy that I do not find in the greyscale mode. Having tested the laser capabilities quite extensively, I decided that perhaps I must be doing something wrong.
I have just used the same settings as are used in the video presentation. The image file I wish to print is currently printing and I will talk about it once it comes off the printer. For absolute certainty, if you were to send me the cat file, tell me the settings which were used and the size to use and if all things are equal, I should be able to produce a file that is similarly detailed.
For now, I will confine myself to discussing what half tone represents. The variation in ink dot can be the size of the dot on the page (analogous to today’s pulse width modulation) or the spacing between the dots (frequency modulation) and it may also be represented by a combination of both techniques. That is to say dot size and space between the dots combined. My example image only demonstrates a change in ink dot size but the spacing between every dot is absolutely regular.
For your half toned laser image I would expect there to be information in the image file that turns the laser on and off at varied intervals (frequency modulation). I would also expect a reproducible pattern to represent every tone available. When selecting which greyscale algorithm I want to use, I assess each one with the image at the same size and then I can see banding artifacts and the smoothness of the digitising process.
The original halftone process required the operator to insert a coarse material screen of between say… 85 and 150 lines per inch (lpi) between the image to be screened and the camera used to image the picture. It would break the path of the light coming through the screen to expose the film by diffraction and halftone printing was the result. The end result is an optical illusion that requires the image to be viewed from a distance where all of the dots coalesce so they are seen by an observer at the correct distance as a continuous tone image.
To aid the clarity I have a couple of images for you to view.
The first image is a screenshot of the file I am currently still printing. I tried to process it in each of the algorithms available for greyscale. The image file is a greyscale image processed at a density of 10 and sized to 95mm square. The algorithm which appeared to give the most pleasing appearance in Luban was Sierra2. (I think that we could usefully have a snapshot facility to compare the algorithms when we are processing images for greyscale)
This first image is how I saw it in Luban… please enlarge the image and tell me what you think the severe banding is between minus 10 to minus thirty. It is actually visible everywhere on the page but between those numbers it is quite severe disruption of the image. It indicates that what is being seen is not the laser software copying halftoning because the man’s head did not and does not have these light bands traversing it. It strongly suggests that the greyscale algorithm is dithering the pixels in an effort to represent tones that are currently out of reach of the software.
Another image may help to clarify this matter for you with more certainty. It is an enlarged section from the first image. It is taken from the centre section of the man’s left eye.
This is a reflection of the combined pulse width and frequency modulation. Tonally speaking, there is very little tonal difference yet I think I see 5 different pulse widths and 2 different frequencies. The regularity is responsible for the banding seen in the previous image and the patterning suggests very strongly that this is a dithering algorithm rather than a halftoning pattern.
The available choice of greyscale image processing algorithms can be summed up like this:
Floyd-Steinberg - dithering algorithm which works by error diffusion.
Jarvis-Judice-Ninke - dithering algorithm works by error diffusion (uses two rows)
Stucki - dithering algorithm works by error diffusion - faster than JJN
Atkinson - dithering algorithm partial error diffusion may wash out contiguous areas
Burke - dithering algorithm works by error diffusion (faster than Stucki and single row)
Sierra 2 - dithering algorithm works by two row error diffusion.
This implies that the enlarged image told me the correct story… that is to say that halftoning was not being applied by the Luban software algorithms for greyscale images. So that leaves me asking for halftoning to be included on the future roadmap and implemented whenever it is possible. It will provide the user with images that appear to be continuous tone rather than dithered.
You may remember that I was exploring potential methods of halftoning images taken into Luban software. I don’t think that my discovery that greyscale images are being dithered (rather than halftoned in Luban) diminishes the value of Snapmaker; but factually speaking image processing is not as convenient as it could be. I was requesting a roadmap insertion that considered the possibility of using the Luban software to drive the laser so as to provide an effect that worked in a manner similar to halftone screening.
David, I sincerely hope that clarifies my position on the point I made about halftones.
The image is still printing and looks to have another 40 minutes to go. I will post it later if I am still awake.