PDA

View Full Version : I think I know what is causing my banding problem - software or firmware bug.



Robert Silvers
10-18-2013, 10:27 PM
Or maybe a corruption in the USB transfer.

I posted some photos that I burned into a mirror, and there was some banding where the laser looked to have become more powerful for a while.

I suspected it was coolant temp, or the power supply changing voltage output. But when a photo turned from a positive to a negative mid-way - that was weird.

I noticed that when I send a job to the printer with a large bitmap that is simple, it transfers quickly. When it is random noise pixels but the same pixel count, it can take up to 12 minutes if full bed size and 300 dpi. This tells me that the software sends bitmaps to the controller ram with run length encoding.

I suspect that either the sending program or the controller firmware is making an error in the encoding or decoding of very complex images.

Both of these photos are the same size - almost 24x36 inches on birth plywood. The more simple image - no problem. The more complex one - very dark band and also went from a positive to a negative mid-way.

What I still need to do is press the START button a second time to re-run the job, and it has the exact same defect, I think that would show with some certainty that the bitmap as stored on the laser controller is corrupt and that the power supply, tube, and cooling - which are analog in nature - are less likely the problem.

http://imageshack.com/scaled/1024x768/34/nz2j.jpg


http://imageshack.com/scaled/1024x768/27/30zx.jpg

Michael Arruda
10-19-2013, 12:43 AM
Have you noticed any groaning when you use the machine? What about blood dripping out of the exhaust port? Did anyone die at the factory the day it was made?

Looking at that photo of the child, my guess is a haunting. You should contact a priest. Or Bob Villa- priest of the Holy Order of Sawdust.

-Michael

PS- or, could be a controller issue. I don't think it would be a transfer problem, more likely a controller not Interpreting the comands properly. Have you tried updating the firmware, reinstalling the driver and software, and swapping to another USB cable?

Bruce Dorworth
10-19-2013, 12:57 AM
I had a very similar thing happen a while back. It turned out to be a mismatch of DPI. I would run a picture through Photograv or one touch. I would tell it to make the image 300dpi. turns out it was 500 dpi. If you have some free time later give this a try. You don't have to run the whole image, just enough to see where it would change.

Hope this helps,
Bruce

Robert Silvers
10-19-2013, 1:13 AM
Thanks Bruce. I am thinking it is probably the super long USB cable it came with. While trying to reproduce the problem, I had times where the transfer just stopped. I didn't measure it, but I have it coiled up like an extension cord. It may be 12-16 feet.

I am making a print now, and I will see what happens. If this one comes out messed up, then I want to save the file from the controller to the USB drive, and then take that file and re-load it back into the PC software and see if the file is corrupt.

Another thing I can try is to load the file into the machine from the USB thumb drive, and try it that way. So there is stuff to test, but of course I am going away for most of next week.

I ordered this cable:

http://www.amazon.com/dp/B004TXPWV6/ref=pe_385040_30332200_pe_309540_26725410_item

Michael Arruda
10-19-2013, 1:49 AM
The usb protocol should be good for 30 feet, but ideally 10 feet or less without a repeater. If the cable it came with is longer than 10 ft and a standard cable, it may well be that. Also check the connector at the laser- mine doesn't make good connection to the cable without some tape holding it together. Oh, the joys of shoddy workmanship on cheap chinese products. But, alas, I am a part of the problem, being a consumer of such things.

-Michael

Robert Silvers
10-19-2013, 10:20 PM
Today I was not able to upload the file at all - it would go super slow - like I could leave it for hours, and it would not be done yet.

I finally figured out that the memory was full. I had a single 35 MB image on the laser controller, and it not only would not load another one, but would not give an error message on the LCD or through the software. They need more ram (they have about 5 cents of memory on it) and more status messages. Same with the copy from USB. I left it "copying" for hours and it never showed a progress indicator or completed.

Kev Williams
10-19-2013, 10:49 PM
What I'm trying figure out is, how did your laser engrave the negative on that wood WHITE?

Robert Silvers
10-19-2013, 10:51 PM
It is no whiter than the wood itself.

Robert Silvers
10-20-2013, 1:55 AM
I loaded a file from the USB thumb drive directly, and it went inverted also. So I guess the USB cable was not the problem. The slow loading was not the cable, and got fast when I clear the controller of all files first. So here is what I know:

1. Controller has very little file storage - probably under 50 MB. Only enough for one large, but not full table size, bitmap.
2. When controller memory get close to full, file transfers slow down dramatically.
3. When controller runs out of space, it does not tell you - it just says "copying" and never finishes.

I went back and read-read the email from the company, and they told me that my file was too big, and maybe that caused errors in the loading process. I don't even remember reading that the first time, so I must have ignored it as implausible. But now I think that is exactly what is happening.

It must render out the image to a native bitmap while loading - so even if I take a tiny jpg and import it, it will still turn it into a huge bitmap if I ask to print it 24x36 inches. Now that I think about it, that makes sense. I had incorrectly assumed that only my original jpeg file size mattered.

They had success printing my file when they output it as 20x30 inches, so that may be the limit.

Rodne Gold
10-20-2013, 3:59 AM
If you want to do pics well using your laser , search for the "gold method" here , or use photograv to process the pic into a one bit image which will be a much smaller file , and it will look a lot better as an engraving.
You then bypass the engravers native greyscale conversion which can be a hit and miss affair , no matter what laser you use.

Robert Silvers
10-20-2013, 12:22 PM
I know that the built in 1-bit conversion has the possibility to be sketchy, so I have been putting the image into Photoshop, making it 1-bit, selecting half-tone, picking that I want the output to be 300dpi, and selecting a line screen.

I have also used the built-in conversion, which actually may be just as good - though I am not sure what it is picking for the final bitmap size that it is generating - though hopefully it is picking whatever is native to the hardware.

In this example - done with the built-in photo conversion, I made the original photo 100 PPI for the final output size. I tried, from left to right, "dot sizes" of 0.51mm, 0.35mm, and 0.25mm.

Those map into line screens of 49.8, 72.57, and 101.6 lines per inch.

Notice how the last one has banding. I believe this is because the 101.6 dots per inch exceeded the original scan of 100 PPI and caused aliasing. The rule is that the original scan should be 1.5-2x the line screen, so I need to retry that with the original at about 200 PPI for max sharpness.

Reference: http://www.autodesk.com/techpubs/aliasstudio/2010/index.html?url=WSa3468f265c3fd43bb1c2601154376c46e-7ff5.htm,topicNumber=d0e52018

"Generally, the pixels per inch of an image should be provided to the printer at a level higher than the line screen (to avoid obvious staircasing and aliasing), but no more than double the line screen (more than double the line screen results in drop-outs, and wasted raster image processing (RIP) for postscript conversion time)."

http://imageshack.com/scaled/1600x1200/41/8pkb.jpg

Rodne Gold
10-20-2013, 2:02 PM
Did you search for the Gold method here?
Doesn't look like it , you seem to eschew advise from those that have been there and done that - it is getting somewhat irritating
Please spare us your theory - there is more to laser photo engraving than just using photoshop to convert to 1 bit... search and and try it and read explanations etc
I assure you it will give you better photoengraving that your current method

Robert Silvers
10-20-2013, 3:20 PM
Did you search for the Gold method here?
Doesn't look like it , you seem to eschew advise from those that have been there and done that - it is getting somewhat irritating
Please spare us your theory - there is more to laser photo engraving than just using photoshop to convert to 1 bit... search and and try it and read explanations etc
I assure you it will give you better photoengraving that your current method

Rodne,

Yes, I saw your method thread. I can't just start out using someone's method without testing it against other ideas and knowing the specific differences.

I made these on MDF - which is not a very high contrast medium. I didn't try to enhance the photo for the medium. I normally would, but my goal was only to compare screen vs diffusion methods and not to make as good an overall image as possible. The physical pieces tell a lot more than seeing these small images on a computer.

Any images that do not say "Photoshop," I did with the PHCad software. The halftone conversion in Photoshop works out better because PHCad converted files have rows of lines that don't appear in the Photoshop version. This is because Photoshop allows you to set the pattern angle to 45 degrees, which the eye does not notice as well. The tonal values between Photoshop screen and PHCad screen were about the same, at least when seen in person.

The lower the line screen value relative to the hardware DPI, the more grays are possible. If I want 256 gray levels, I need to print at 16 times as high DPI as my line screen. So if I print at 600x600 DPI, then I would need a line screen of 37.5. That would not be as sharp as 75 LPI, but 75 could only reproduce brightness 64 levels. 300x300 DPI printing will be even more limited. Which screen I pick will depend on the situation.

I tried diffusion at 100, 150, 300, and 600 DPI, and it gets darker, so I have to work out the best way to adjust for that.

Gold Method - the sharpening helped make the hair pop, but 500% was too much for even the area that it was helping with. It totally destroyed the skin tone values, while also adding ringing to the edges of features. At the moment, I just want more of a full tonal range look for my goal. If you want to put someone's face on an award, it should make the customer happy as they will appreciate the vibrance.

My conclusion - with my printer, and the full-tonal-range look I am going for, I will probably use Photoshop with a line screen in the 50-75 range, depending on if I prefer spacial resolution or better gray levels - or work out a way to best adjust for the darkening of the diffusion. If I want to make my image pop, I can adjust levels and sharpen as needed - but I would tend to mask the image and do it only in certain areas. In this case, I would do it for the hair, background, and shirt, but not the face.




http://imageshack.com/a/img22/8317/3qx9.jpg

Jeff Woodcock
10-20-2013, 5:11 PM
I prefer using the Woodcock 8bit Method. :)

Scott Shepherd
10-20-2013, 5:21 PM
I'm not overly impressed with any of them.

Gary Hair
10-20-2013, 5:21 PM
The 200ppi source and the Gold Method are the better results, in my opinion. What you want out of the picture determines which you prefer. For most of what my customers want, the Gold Method would be the better option. That said, I use PhotoGrav and get results similar to the Gold Method.

Robert Silvers
10-20-2013, 5:49 PM
I'm not overly impressed with any of them.

My goal was to compare line screens at various DPIs with each other and with error-diffusion. I can work on other issues like shadow detail now.

Bruce Dorworth
10-21-2013, 3:14 AM
Robert, is this not what I said a few days ago? That it was an issue with the DPI.

I guess if you have your own ideas of what is causing your problem, then you should exhaust those option before asking for advice.

Bruce

Rodne Gold
10-21-2013, 4:29 AM
A laser is not a a photo printer - that it can do pics is a huge bonus , but you ain't gonna get what you would out of a cheap 300dpi laser printer.
We use photgrav for pics , no fiddling and they come out great on whatever substrate (Gold method basically mimics what Photograv does)
But then again , we aren't tinkers.. I have a commercial operation so need to do it right 1st time.

Robert Silvers
10-21-2013, 6:49 AM
Robert, is this not what I said a few days ago? That it was an issue with the DPI.

I guess if you have your own ideas of what is causing your problem, then you should exhaust those option before asking for advice.

Bruce

The machine ran out of memory without giving an error. I can't find where you said that was a potential problem. If you did, let me know. The manufacturer told me that after I asked them what it could be, and my tests verified it. This thread being here may help the next person.

Robert Silvers
10-21-2013, 9:46 AM
A laser is not a a photo printer - that it can do pics is a huge bonus , but you ain't gonna get what you would out of a cheap 300dpi laser printer.

Yeah. Last night I used my laser printer to compare line screens to diffusion dither. It was interesting. At 300 DPI the 100 line screen had serious quantizing error. The 75 was not bad. I liked the 50-60 the best - which is actually what 300 DPI laser printers tend to use.

The diffusion method was ok after I lightened the source image, but it was very noisy - and I can't say that I preferred it.

It appears that most of the marketing efforts touting frequency-modulated half-toning methods only compare them to "standard" line-screen sizes and then use a press with high resolution to show a how an FM image is better. What they don't do is use the highest screen that the printer was capable of. That is misleading but seems to be unnoticed.