PDA

View Full Version : Huge files



Joe Pelonio
07-10-2006, 5:52 PM
I'm engraving photos and text onto annodized aluminum, and the customer scanned the photos at 1200 dpi, so the file size of the first one is 89 MB.
It took about 15 minutes to get from the PC to the laser. I'm not going to buy a new computer for one job, but is there something like more memory or a faster video card that I can put in to speed it up? It's a Dell 4600 pentium 586 2.4 Ghz, with math coprocessor, 256 MB Ram, running XP, hardwired direct network connection to laser.

Bob Tate
07-10-2006, 6:03 PM
XP doesn't even seem alive until it has 512. Memory is the biggest performance contributor. After you have memory then the bottleneck is Video.

Joe Pelonio
07-10-2006, 6:12 PM
Thanks Bob.

Jeff Chumbley
07-10-2006, 6:45 PM
Joe RAM is cheap get at least a gig


Jeff

Joe Pelonio
07-10-2006, 7:01 PM
Looks like it has to be installed in pairs so I'll get 2 x 512 MB. Thanks.

Jeff Chumbley
07-10-2006, 7:17 PM
Sorry just noticed its a 586 yes its pairs on EDO.

Jeff

Dave Jones
07-10-2006, 7:52 PM
I've been doing a lot of anodized aluminum, and with my Epilog Mini-24 / 45 watt I found that the best results came at 600dpi. That would cut your file size 4:1 from 1200dpi.

Of course different lasers will be different.

James Stokes
07-11-2006, 7:06 AM
Very seldom do I run anything above 400 dpi. I have found that to be good for anything I want to engrave. With that I can engrave text so small you have to use a magnifying glass to read it. For photos I try to keep those to 200dpi.

Rodne Gold
07-11-2006, 7:22 AM
At best , the laser can resolve 150 ppi , so the formula for scanning is this:
DPI = output size/input size (in inches) * 150
IE if you have a 6" x 4" pic and are going to engrave it as a 12x8" the DPI you need for scanning is 12/6X150 - IE 300dpi
1200 DPI scans are stupid unless you plan to enlarge the scanned pic real big!!
Trying to run XP on 256 mb ram is a futile excercise , 1-2 gigs is the minimum , however it wont help with files the size you are talking about as these will page to disc big time , you need something like a raid array to deal with such stuff. We run huge digital printers where we regularily deal with such big files and worse , disc performance and processor performance are king here.

Bruce Volden
07-11-2006, 7:23 AM
Joe

I appreciate it when customers use higher resolution!! Then I can run it through PhotoPaint and resample it to deal with file size. No whining~~if I had to dump a file that large to my old LMI (serial port) I could have went out for steak!!!???


Bruce

Joe Pelonio
07-11-2006, 7:54 AM
In this case the photos are scans of pics taken in the early 1900s, so the
clarity was not too good to begin with, then old, so they are scanning at 1200 to try to keep as much as possible of the detail. I'm planning to experiment but will probably end up engraving at 600.

Matt Meiser
07-11-2006, 8:15 AM
I know nothing about lasers, but you mentioned transfering the file from the PC to the laser across a network connection. What is the speed of that connection? If it is a 10 megabit network and you can increase that to 100 megabit or even 1 gigabit, that would significantly improve the time transfering the file across the network. 100 megabit network cards can be had dirt cheap these days.

Like I said though, I don't know anything about lasers, so maybe that's not the issue.

David Fairfield
07-11-2006, 11:21 AM
In this case the photos are scans of pics taken in the early 1900s, so the
clarity was not too good to begin with, then old, so they are scanning at 1200 to try to keep as much as possible of the detail. I'm planning to experiment but will probably end up engraving at 600.

Bit of a tangent here, but glass plate photography in the early 1900s was magnificent quality, better than any 35mm film photography I've seen.
I collect old photos, and many of these turn-of-the-century prints is like looking into a crystal ball, perfectly sharp down to the minutest detail. What you probably have is a scan of a reprint. Any chance your client has original prints or the glass negatives?

Dave

Joe Pelonio
07-11-2006, 12:04 PM
The client providing the files is a city, and the photos from their historical archives. I don't know what kind of originals they have, but when they scanned it at a lower resolution they were really bad. I'm guessing that they are scanned copies of originals, but these are for an exhibit and they want it to look as good as possible. Unfortunately everything is through another party, as this job comes from one of my wholesale customers.

David Fairfield
07-11-2006, 1:10 PM
City and museum archives are really good sources for original prints. If you think the scan is less than ideal quality, maybe mention to the client you could achieve better results with a better original, should one be available.

It seems obvious but... I've noticed over the years that many otherwise intelligent people can be dense when it comes to doing something outside their area of experience. So they might assume for example, that working from a copy is as good as working from an original. Or they'll just give you a copy because they had it handy, instead of the original, then wonder why the final product image isn't sharp.

Dave

Dave Jones
07-11-2006, 2:16 PM
At best , the laser can resolve 150 ppi

Not true with my laser (Epilog Mini-24), and I don't have any special lenses or anything. I did a series of tests on black anodized aluminum and at 150 dpi there was a clear gap between dots. The size of the dot changed with power settings, and at what I felt was a good bright mark on black anodize required 300dpi before the gaps were gone between dots, and required 600dpi before the jaggies smoothed out.

Your visual resolution limit is not at the point where your dots touch each other. It's well beyond that. It's more like where the dots are at least 50% overlapping. Individual dots give higher perceived resolution when they can be positioned on a finer grid, and multiple overlapping dots give smoother, sharper edges to lines and curves.

In my tests, which used images of 300, 400, and 600 ppi, printed at 150, 200, 300, 400, 600 and 1200 dpi on my Epilog onto black anodized aluminum, I feel the best results came from a 300ppi image printed at a 600dpi setting in the driver.

Michael McDuffie
07-11-2006, 6:55 PM
Joe,

I assume you are using IntelliScribe and the XP driver. I don't know about your machine but my Legend24 has a 10Mb ethernet connection which is quite slow these days.

If you watch the log as it transfers, you will see that IntelliScribe sends a chunk, waits for a reply then sends another chunk and so on. I assume the machine takes those chunks and reassembles them as it goes and I don't think the processor is all that fast.

I have a file that takes up to ten minutes to transfer!

You might try setting the driver for 600DPI and running a test piece. That should cut the transfer time a lot and the engraving time just about in half.

Michael

Joe Pelonio
07-11-2006, 7:46 PM
That sounds like the way it works, I send the file from Corel 12 and get the bar graph, very slowly filling up. When the job appears on the laser I can view it and parts are missing. View it again and more is there. I found that when rastoring I can go ahead and start it and the rest will be there in time. I'm waiting for the aluminum to come in, then I'll try a detailed part of the photo at 1200 and again at 600 to see if it's enough difference to be worth the extra time.

Rodne Gold
07-12-2006, 1:09 AM
Dave , the resolution you engrave at is totally irrelevant in terms of what I am talking of , there is huge confusion here between PPI and DPI .

My 1440 DPI large format printers cannot resolve any more than 175 -200 pixels per inch. A pixel in a photo is either a shade of grey or a colour , lets take the lasers case for this example.

How many "colours" can a laser engrave? It can only do 2 colours , either no engraving or a mark. You cannot represent anything barring black or white this way. So in essence a "dot" can either be black or white , however a PIXEL can be anything inbetween.

For the laser to represent a shade of grey , it has to fool the eye and the way it does so is to represent a pixel as a matrix of dots , lets say a 3x3 matrix ... White would have NO engraving in the 9 cell matrix , Black would have ALL the cells of that matrix coloured in , lightest grey would have one cell coloured in etc etc.

These Cells , not the dots, are the pixels and thus define the resolution. IE if your laser is firing at 300 dpi and you use a 3x3 cell matrix , the BEST you can represent is 100 pixels per inch. We can take this further too and say with a 2" lens bearing its spot size in mind , you cannot actually print better than 300 DOTS per inch cos as you exceed that , the spots the laser makes are no longer discrete and run into each other.

Perhaps on your laser with your motion system and on the particular media with its resolving power and dot gain with your spot size and power level, you got the best results using those ppi numbers , but its particular to that application and is not a generic thing. You cannot extrapolate it. There is a case to be made for scanning at double the PPI that the device can resolve (hence you finding that 300ppi images were best) and that is from Nyquists theorum that states that to perfectly represent an event , it has to be sampled at 2x its frequency. This however has empirically been proven to be overkill (in terms of printing) and a figure of 1.25 is mostly used.

I still stand by the statement that a laser cannot resolve better than 150PPI if you want shades of grey , its a physical impossibility , thus scanning at 1200ppi is just a waste of time and resources as has been proven here , a total loss of productivity (15 mins to transfer a file is rediculous)

Why on earth would one want a file that one has to resample downwards and thus lose info on (downsampling is more destructive than scanning at the correct PPI)
Why would you want something that is going to slow your system down and give you no benefit whatsoever?

I do not only do Laser "printing" , I run a few large format digital printers (1440dpi machines - 1.6m wide) , multiple a3 photo inkjets , thermal resin printers and so forth and these principles apply to all of them.

Shaddy Dedmore
07-12-2006, 4:16 AM
I'm just trying to keep up, so I have a question.

So you're not arguing that setting the laser at higher than 150dpi is a waste, only that scanning in the photo to begin with is?


I know that on a good piece of marble (not lasersketch) I've seen noticable differences in 300, 600 and 1200 dpi settings for the same photo (that was scanned in at 300dpi). So I think you're refering to the photo, not the laser settings.

So if I'm understanding, the test to verify what you said would be to scan in a photo at 300 dpi and run it at those settings, then compare it to a photo scanned in at 150 dpi at the same laser settings.

But as for transferring to the laser, I think sending a 300dpi photo at a laser setting of 1200 is the same as sending a 1200dpi photo at 1200.

I'm just trying to get a handle on things, thanks.

Shaddy

Rodne Gold
07-12-2006, 10:38 AM
DPI in terms of engraving is not really related to scan DPI , you might find engraving at X Dpi better than 1/2X or 2xX , but this has little to do with the amount of pixels per inch you are engraving.
The DPI you engrave at has a HUGE amount of variables to consider , not only spot size , but the depth of engraving , the dot "gain" (IE the heat affected zone around the vaporised "dot" ) How the laser pulses , when it pulses , the accuracy of the motion system and the accuracy of the sensors that tell it where it is , ramping , the surface of the media , focus issues , flatness of the media , media movement and so on. It's impossible to say to anyone what the best DPI to engrave is.
An image scanned at 1200ppi will be 16x bigger than a 300ppi and the driver will have to do some serious maths if its going to be engraved at 300 dpi. It has to take a whole lot of extra scanned pixels and try to combine them to 1 and will have to then guestimate a shade of grey using the various shades of grey/colour of the pixels it combined. A huge amount of overhead for nothing. To Manipulate data like that , one uses whats called a RIP (Raster image processor) and these can cost big money as well as requiring superfast computers. A Rip would also have to have profiles for various output devices and to make these requires all sorts of jumping thru hoops.
Sending a 300ppi file at 1200dpi to the laser is very different , the laser can actually do a better job than the other way round as it has a 4x4 dot matrix to represent one pixel , about which it knows everything.
At the end of it all , the laser drivers and programs like Photograv are VERY simplistic devices and relying on them to do the correct manipulation to get best output is sometimes a hit and miss affair and it's something one has to accept.

Dave Jones
07-12-2006, 3:43 PM
Rodney (is it Rodne or Rodney?),
What you are saying is absolutely true if you are using halftone conversion, which is grid based. But it's only partly true if using diffusion dithering or noise dithering. While those methods can not convert every pixel, they can give much cleaner edges to strong transitions when the input image is full resolution.

In other words, if you start with 100ppi image and use halftone conversion to convert it to a 300ppi bitmap, and then engrave that on the laser at 600dpi, this is as good as you can do with a halftone process.

But you will get better results by starting with a 300ppi image and converting it to a 300ppi bitmap with diffusion dither. It still won't resolve individual pixels below 100ppi, if they are similar shades to their neighbors, but it will give a much less jagged line to edges where the shades transition with decent contrast. That's because diffusion and noise dithered images don't stick with a grid. Those sharper edges and lines will engrave dots on the full 300ppi grid, and having that extra positioning of the dots will make transitions in the image much clearer than with a 100ppi image converted with a halftone.

So for most photographs I firmly believe you will get the best results by feeding a 300ppi image to Photgrav (or do the dithering in Photoshop yourself) and getting back a 300ppi bitmap. Then engrave that at 600dpi on the laser.

To prove my point I have created the following graphic. It shows a one inch portion of a grayscale image at 300ppi. I convert it to a 300ppi bitmap using diffusion dither. I also convert it to a 100ppi greyscale image and then to a 300ppi halftone bitmap using a 100 line screen (which gives the 3x3 cells you mentioned). For the fun of it I also converted the 300ppi grayscale image directly to a 300ppi halftone bitmap using the same 100 line screen and was surprised that even that is slightly sharper that the 100ppi original converted to the same halftone. (this is all using Photoshop). The final row on the graphic takes those 300ppi bitmaps and gives them a 1 pixel gausian blur and a 30% size reduction to simulate roughly what you would see when you stand back from the final engraving.

This simulation is all based on sending the exact same resolution bitmaps to the laser so this is not about what settings work best on the laser. The exact same principles are true no matter what the final bitmap resolution is.

http://www.redtengu.com/images/raw/laser_ppi.gif


As far as oversampling the scan, oversampling and then shrinking will eliminate aliasing caused by thin lines or details in the original. If the scanner simply samples points at a lower resolution it can end up sampling small points that cause aliasing artifacts in the scan, while oversampling and reducing will tend to average those fine points together (assuming you use a downsampling like Bicubic rather than Nearest Neighbor).

Also oversampling and shrinking using a process like Bicubic Sharper in Photoshop will result in a slightly enhanced edge sharpness over an image scanned directly at the final resolution.

Dave Jones
07-12-2006, 3:49 PM
I forgot to also say, everything I mentioned above about diffusion dither applies also to Photograv if you use it in dithering mode. There's not a lot of difference between diffusion dither and noise dither, at least when comparing them to halftoning. Photograv has a bit more control over the dithering process than Photoshop, so generally will give even better results than what I showed in my sample image above.

Rodne Gold
07-13-2006, 1:33 AM
Dave , it's Rodney:)
I was only using the grid type agorithym to demonstrate the difference between DPI in terms of engraving and its realtionship to pixels per inch in the graphic. IE one pixel does not equal one dot when printing.

I had said in theory , one should scan at 2x the output resolution , so assuming 150ppi max resolution of the laser, I agree with you in that 300ppi is about as far as one should go if the image isnt to be enlarged.

Despite saying photograv is somewhat simplistic , My opinion is that if you do want to get into engraving photos , Photograv is a doddle to use and is a no brainer choice especially if one has only one chance to get it right as the results will be acceptable almost 99% of the time. It can also be used by relatively unskilled operators. We use it for a lot of other stuff other than photos , like for engraving complex vector clipart where it works far better than our laser drivers conversion.

Dave Jones
07-13-2006, 8:54 AM
Rodney, I thought I'd seen a "y" on the end of your name on other forums. That's why I asked. :D

I agree that there's no point going past a 300ppi grayscale image to send to the laser. Your initial post in this thread had said 150, which was what I was arguing against.

I do still feel there is an advantage to scanning higher than that and reducing down to 300ppi to reduce aliasing artifacts. At least for those of us trying to squeeze out every last bit of quality in the image. It's subtle, but worth it to me.

I've played a bit with Photograv and think it is brilliant. It's defaults are good, but if you tweak it's settings in interactive mode you can get even better results. Photoshop can give similar results, but it's a lot more work and Photoshop's conversion to diffusion dither has a bug that creates a sharp edge along the top of the dithering.

I don't know about other lasers, but with my Epilog the built in halftoning is nowhere near as good as using Photograv.

Joe Pelonio
07-17-2006, 7:20 PM
I know this thread evolved to an interesting discussion of resolution, but to go back for a minute to my initial question, I just installed another 512
of memory. Sending the same 89MB file that took 15 minutes before took
4:32, a big improvement. :cool: Thanks for the help Bob & Jeff.

As for the resolution issue, still waiting for the anodized aluminum to arrive.

Dave Carey
07-18-2006, 1:01 PM
Joe,
I don't know if this is germaine or not (and it may not help the transfer of large files once you've received them) but Walt Mossberg, who does the personal tech column for the Wall Street Journal, favorably reviewed software that helps move large files from user to user. The article was last week (Thursday I think) and is probably available from the WSJ website. The company was www.pando.com and the software (Beta version) is free and available for windows and a mac. Dave

Jeff Chumbley
07-19-2006, 9:07 PM
Its like buying a truck. You always!!!!! Need more power. The higher the processer and the more RAM the better. I have just upgraded to a gig. I am building another desktop (or should I say attempting to) and plan on 2 gig with plenty of disk space. It's so cheap compared to the "old days" you can't not do it.

Jeff