I’m using CS.
I’ve found that the print option is grayed out when I create a image where one of the dimensions is larger than 30,000 pixels.
Is there a way to print and image this large?
Also, I’ve heard that Epson has a limit as to the length of a print. Anybody know about this?
Thanks!
Steve
In many image formats, the dimensions are stored as signed 16-bit integers; in such formats, the dimension could be any number of pixels from 0 to 32,767, meaning the maximum number of distinct pixel values would be 32,768. I believe Photoshop itself has this limit for image dimensions (sometimes expressed as "30,000 pixels", which is "close enough for government work") unless large image support has been enabled (and such files would have to be saved as PSBs).
I have no doubt that this limit, as to printing, is due to the likelihood that some or all printer drivers will not handle larger numbers of pixel dimensions properly. While the Windows XP graphics display interface uses 32-bit variables for bitmap dimensions, I’m sure earlier OS versions were limited to 16 bits, and there may still be drivers out there that can’t properly handle images with dimensions exceeding 32,768 pixels.
As far as I know, the only way to print an image > 32,767 pixels in any given dimension from Photoshop is to resize it so it doesn’t exceed the max. You can use software known as a RIP, or raster image processor, to print images that exceed this size limit. I haven’t used such software, so I don’t know whether the programs work within Photoshop or require printing via a separate interface.
As to Epson, there is definitely a hard-coded length limit in the driver. It’s supposedly 44", which corresponds roughly to an unsigned 16-bit integer (65536 values) at 1440 dots per inch (i.e., 65536/1440 just happens to be 45.531"). This can be overcome using a RIP. I have done 40" panos but no larger on my Epson, so I haven’t run into any problems.
This may also have governed Windows bitmap dimensions in the past (I believe that in XP the OS uses 32-bit integers), and some graphics drivers may have such limits hard-coded in them.
Another way of approaching this is to consider the resolution of the print. We habitually tend to produce files for printing at 300ppi. 30,000 pixels at this resolution will give you a 100 inch long print.
No print even approaching that size is going to be viewed close up. So when you want a longer print, simply resize without resampling. Doubling the size (in this example) to 200 inches long would result in a 150ppi file. Ample resolution for a poster.
Chris.
The OS print APIs use signed 16 bit numbers.
So Photoshop limits printing to 30,000 pixels (leaving a little leeway).
Chris,
You actually left more than a little leeway. Since 215 = 32,768, you could have set the limit at 32,700 pixels and still you would have had over 60 pixels of leeway. That extra 2,700 pixels that we didn’t get is actually quite a bit of image, particularly when you calculate what it entails area-wise. Let’s do the math: We could have had 32,700 x 32,700 = 1,069,290,000 pixels. What we got was 30,000 x 30,000 = 900,000,000 pixels. So we got shorted at least 169,290,000 pixels. I wouldn’t call that "a little" leeway. (grin)
— Burton —
The leeway is not just for pixels, it’s also for headers and other things needed when you save a file.
Cybernetic Nomad,
It doesn’t seem logical to me that the 16-bit number limitation would have anything to do with the size of the headers and other info included in the file. In fact, I am not certain I agree that there needs to be any leeway at all.
I’m also curious why a signed integer is being used to address the pixels. I would think that the pixel addresses could range from zero to some maximum positive number. I don’t see the need for negative pixel addresses. That’s a little bit like a negative house number or a negative phone number. We may have been shorted by a factor of two in linear dimensions or by a factor of four in image area pixels.
I do recognize that many image calculations do need to extend the image beyond its borders to avoid undesirable effects near the borders. Perhaps that requires a few pixels of leeway, but a leeway of over 2,700 pixels seems very excessive to me. And, considering the sign bit thing, the current leeway may be over 32,700 pixels.
— Burton —
Since the Pre-CS pixel limit was 30,000 pixels – we just left the limits there.
Ahnd you need signed values to represent things off to the top left (including padding for filters, brushes, etc. and the position on the page). And the OS also uses signed values for the same reason.
2700 is only 9% of the 30,000 pixels that are allowed to print. And from experience, it’s better to leave a little space for the OS to do other things with the coordinates.
Chris,
Well, I respect your authority as one of the senior programmers and you certainly have good knowledge of the internals of Photoshop while I know nothing in that area. And I recognize that Photoshop’s software engineers have more pressing things to do than re-optimize a file structure design that is working pretty well.
But that said, I think that at some time in the not too distant future the 30,000 pixel limit should be readdressed. When that limit was originally set, hard drive capacities were a small fraction of what they are now and commonly available RAM capacities were also a small fraction of today’s RAMs.
As I said, I recognize that filters and brushes need to extend the image somewhat in order to treat the pixels at the borders of the image with the same algorithms as interior pixels. And that does mean that there need to be pixels to the left of and above the upper left corner of the image. The image extensions go all around the image such that pixels exist below and to the right of the lower righthand corner of the image.
However, you can prevent that situation from forcing you to use wasteful signed integer addresses for pixels by simply moving the origin from the upper left corner to a point above and to the left of the upper left corner by enough to provide your extrapolated image extensions.
As to the role of the OS in this, surely the OS allows the use of both signed integers and unsigned integers. Those guys dealing with aerial photography, or satellite images of weather patterns or the earth’s surface, need to work with some pretty big bitmaps.
— Burton —
I think that at some time in the not too distant future the 30,000 pixel limit should be readdressed
You’re a bit late there <g> with CS the limit is 300 000 pixels (ten times as many!) and a brand new file format was created to save them: PSB
We just have to wait for the rest of the software world to catch up…
Yeah, saving a truly massive .psb is great!
It would be greater if you could actually DO anything with one.
John,
It would be greater if you could actually DO anything with one.
Can you print one? If so, the little discussion I have been having about the 30,000-pixel limit would seem to be kind of moot.
— Burton —
My point exactly, I think.
Burton – well, first the OSes have to allow you to print something larger. They’ve added some new printing APIs recently, but they’re not that well tested as of yet.
Chris,
Well, I am hoping that the testing of the new printing APIs is well underway, so that this problem will go away with the next Photoshop release. It seems like we have been discussing this 30,000-pixel limit ever since I can remember. Perhaps we are approaching a tipping point where the number of manhours spent discussing the limit will exceed the number of manhours needed to remove it. (grin)
— Burton —
Unfortunately, no.
The 30,000 pixel limit was removable for documents – but printing and certain file formats are outside of our direct control.
Since going beyond that limit means completely changing print APIs, it will not be a trivial task to make the change.
