The most technologically advanced drink-coaster?

June 21, 2010 3:29 PM

I would like to etch interesting visible patterns into CD-R or CD-RW (or whatever disk media). Sort of like those old "spirograph" toy patterns. I would like to use a standard CD drive, etching a visible pattern with that laser. Is this even possible? How could it be possible?

Ultimately, I'd use them as rather interesting drink coasters. Or Christmas tree ornaments. Or whatever. The goal is purely visual, no data need be readable.

posted by garfy3 to computers & internet (13 answers total)

You might look into LightScribe discs.
posted by Blazecock Pileon at 3:35 PM on June 21, 2010

Lightscribe CDs/DVDs have you flip the disc over and burn a label into the top using the drive's laser. You need both a compatible drive and media.
posted by GuyZero at 3:37 PM on June 21, 2010

maybe look into LightScribe media and drives - you can usually pick up the discs at most office-type supply stores, and a lot of PCs nowadays have LightScribe DVD drives (or you can get an external/replacement internal one that does it). it doesn't really etch, though. it's pretty similar to how CD burning works (you're really just making a dark spot in some ink). the lasers in CD drives aren't really at all powerful enough for actual etching. maybe if you removed the laser and put a needle or something else pointy in there instead..
posted by mrg at 3:37 PM on June 21, 2010

On a completely random note, have you tried putting a CD-R into your microwave for 1-3 seconds? If all you're after is interesting visuals, you'll get that. Very interesting spiderweb patterns, especially with those "neon" colored discs.

Usual "putting things in microwave" disclaimer applies.
posted by xedrik at 3:57 PM on June 21, 2010 [2 favorites]

It should be totally possible to make a pattern appear on a CD without specialized equipment. CD's look different when they have data or if they don't, so it's simply a matter of figuring out a mapping from xy co-ordinates to data-space and then writing data wherever you want "ink". Though I don't know how good an image you would be able to produce. I thought I saw someone doing this on BoingBoing but couldn't find it. Here's one attempt I was able to find.
posted by PercussivePaul at 4:00 PM on June 21, 2010

I think garfy3 wants to burn an interesting pattern with the "data" portion of the disk.

When you burn a disk half-full of data, it is obvious where the burn started and where it ends. I'm not sure if you can actually distinguish between a large region of 0-data and 1-data, or if you can just see where the laser has modified the pigments on a CDR.

The first experiment to try is to cook up a CD image that consists of two binary files - one with 300MB of 0s and then one with 300MB of 1s. Burn that and see if the result looks interesting. If it does, then you should have some insight in to how to create your cool image file.

If not, then it could be that CD data is encoded in some non-literal way. Network wire-protocols sometimes encoded data so that the distribution of 1s and 0s ends up with a distribution close to 50-50 to ensure that no large charges end up on the wire, so it is possible that CD data uses a similar process.
posted by b1tr0t at 4:02 PM on June 21, 2010

The feature your looking for is not Lightscribe which puts a pattern onthe top or label side. Your looking for T@2 which puts a pattern on the data or bottom side. It's not well supported so good luck actually finding a compatible drive and media.
posted by Confess, Fletch at 4:44 PM on June 21, 2010

It should be possible. My first instinct is to say that you probably want to bypass everything and write directly to the device, otherwise you'll get lots of metadata and other noise that you don't really need. That won't solve all your problems, however, because the CD recorder itself is going to encode stuff in a particular way to make it readable. I believe that 1s and 0s aren't encoded directly, but rather a pit will indicate "same as previous value" or "differs from previous value". There are also rules about how many identical values can appear in a row and error coding on top of that and I doubt that is done with the device driver.

I'm not sure if you can bypass any of that nonsense, so you might have to find the right mode to minimize that amount of error coding. Then start writing interesting patters to the disk. 0x00, 0xff, 0x55 (alternating bits) as examples. Write, oh... I don't know... 50MB of each of these to the disk and then see what it looks like.

My vague recollection is that each track on a CD holds the same amount of data (meaning that it's squished at the center and more spread out at the egdes) and that this is not true for DVDs, but I could be competely off here.

I think it's time to buy the cheapest set of CDR disks you can find and experiment.
posted by It's Never Lurgi at 5:03 PM on June 21, 2010

I don't think this will work. The raw data is encoded with extra ECC bits as well as framing bits so even if you write all zeros at the application level it will still be a mixture of 1s and 0s at the disc level. And it's not like you can just seek to a random track and start writing there: the tracking optics need the framing bits in order to lock in on the data and determine what track is being read, so everything on the disc must be in a sequential linear pattern without gaps.
posted by Rhomboid at 5:55 PM on June 21, 2010 [1 favorite]

I'd experiment with CDRW discs. Try different bit patterns & lengths.
posted by Pronoiac at 6:31 PM on June 21, 2010

I have a Yamaha CRW-F1 CD burner, which is supposed to be one of the best Yamaha ever made...before they got out of the CD-burner business altogether. It uses the T@2 system Confess, Fletch mentions. There's a little graphics program where you create your art (and it shows the amount of space you have left to work with after burning your data). The result is not really very noticeable, but it's supposed to show up better with those blue-toned CD-RWs.
posted by Bron at 7:53 PM on June 21, 2010

The encoding of data bits onto a CD is fairly elaborate— lots of interleaving, two separate Reed-Solomon coding steps, 8B/14B modulation, sentinels, sync marks, blah blah. It would not be too hard in theory to reverse the process and go from a desired image to a data file which, when encoded, would approximate that image, but I don't know if you'd really be able to get rid of all the synchronization words and so on. Writing file with long runs of 0s and 1s will almost certainly not produce interesting visual patterns though, they'll all get randomized in the name of error robustness.

(I assume that the DiskT@2 and similar technologies bypass the fiction of data coding and simply send image data directly to the laser.)
posted by hattifattener at 1:30 AM on June 22, 2010

My vague recollection is that each track on a CD holds the same amount of data (meaning that it's squished at the center and more spread out at the egdes) and that this is not true for DVDs, but I could be competely off here.

You are completely off here. First of all, CDs have a single spiral track of data, like a record, rather than cylinders like hard drives (which I assume is what you mean by tracks - The "tracks" that your cd player shows you are just some (up to 99) mostly-arbitrary offsets into that spiral listed in the table of contents at the beginning, which is in the middle). Secondly, CDs at 1x are CLV (constant linear velocity) not CAV (constant angular velocity). This is why you can't skip quickly on a standard cd player - the spindle has to speed up or down as you move in or out. But a "48X" drive is typically CAV, and only achieves 48X the datarate at the outer rim of the disc - in the middle it's only reading 18x as fast as a 1x cd. This basically means that the pits and lands are as small as they can make them everywhere on the disc (otherwise they'd have to be 250% as large on the outside as on the inside!) But it means you can't seek that quickly, and it's a little harder to figure out where physically the 180 millionth byte goes.
posted by aubilenon at 1:51 AM on June 22, 2010