Merging multiple images into a single image

[theLoster]

I'm not trying to do photomerging where you combine two separate halves of an image together, but rather I am trying to merge a whole bunch of smaller (tiles) together into a single big image.
If you look at the page at http://tittiesondisplay.com/PBLinks/...hp?y=1979&m=08 it has a whole bunch of smaller images stuck together to form one high resolution centerf0ld. I don't need to do any kind of color matching or alignment as all the various smaller images already fit together. I just need some way to merge them into a single image so I don't have 500 smaller images per picture.

[pigulici]

I think you need some script that combine the files looking at the names, because the names look like x and y coordinates, to me...

LE:
See this if you have Photoshop:
https://forums.adobe.com/thread/492005

or this(not tested by me):
https://github.com/Markavian/tile-merger

[retroanalyst]

You really want something that can be scripted here, nobody wants to work on 560 files with mouse clicks. Fortunately that's not too difficult since the big picture is nicely grouped into a grid of 35x16 small images, each one 256x256 pixels (except for the last row which is only 211 pixels high).

The convert command from the ImageMagick suite can stitch these together, I did it the following way (this assumes a shell like bash):

for column in {0..15}; do convert -append 6-$column-{0..34}.jpg ${column}.jpg; done
convert +append {0..15}.jpg Dorothy_Stratten.jpg

First we combine all the rows, then the colums. The result:



Enjoy!

[theLoster]

Thanks for the heads up on ImageMagick. I've been playing around with it in php. I've been able to assemble various tiles together into a single image. The code I have for displaying it is

  header("Content-Type: image/png"); 
  $combined->setImageFormat("png");
  echo $combined;

The thing is, with that, the image is the only thing I can have on the webpage. Might you (or anyone) know how I can embed the contents of an ImageMagick image object as an html IMG object. I'd almost want to put something like

echo " <img alt=\"Embedded Image\" src=\"";
echo $combined;
echo "\">";

But I obviously have to do more to it than that since all I get is page after page of hieroglyphics.

(disclaimer: I don't know a whole lot about scripting, but I'm pretty good and copying examples and beating on it till I'm able to force it to do what I want)

[theLoster]

A little more internet searching found the trick for me. The line I ended up with is

echo "<img src='data:image/png;base64,".base64_encode($combined)."' />";

And that did it for me.

I'm probably not going to modify the script that I have at the above provided link (sorry). I'd be too afraid that someone might get the impression that I'm hosting the images rather than merely linking back to playb0y for them.

I think I will put a link on the index page for them back to here to show others how they might be able to do it (if you don't mind).

[yag-kosha]

Originally Posted by theLoster

A little more internet searching found the trick for me. The line I ended up with is

echo "<img src='data:image/png;base64,".base64_encode($combined)."' />";

And that did it for me.

I'm probably not going to modify the script that I have at the above provided link (sorry). I'd be too afraid that someone might get the impression that I'm hosting the images rather than merely linking back to playb0y for them.

I think I will put a link on the index page for them back to here to show others how they might be able to do it (if you don't mind).

Is there a (brief) tutorial for those of us mentally-stunted types who aren't programmers or coders? When I click on one of the links I get nothing but alphanumeric data -- no pix!

[theLoster]

What's described here would require at least php programming. Also, the individual image tiles on the megazoom page used to be freely available to download from the rabbit company's website, but they are no longer available (at least that I've been able to find).

If you have the patience, you can find some of them still available from the wayback machine (via the link you included). I just checked one and it wasn't there, then checked another and it was, so it seems to be a hit or miss proposition.

That alphanumeric stuff you're seeing is the indication that the image tiles are no longer available.

[premu]

Sorry for reviving a rather old thread, but the web page in question still links to here, and the info in this thread still contains useful and relevant information. I'll just comment on a couple of points which might be useful for anyone else trying to do the same (kind of) thing.

Originally Posted by theLoster

Also, the individual image tiles on the megazoom page used to be freely available to download from the rabbit company's website, but they are no longer available (at least that I've been able to find).

If you have the patience, you can find some of them still available from the wayback machine (via the link you included). I just checked one and it wasn't there, then checked another and it was, so it seems to be a hit or miss proposition.

It is a bit hit and miss indeed. I did a few checks myself (by no means comprehensive), and it seems that all centerf0ld tiles on this site from Miss(es) January 2000 onwards are in the Archive, but there is nothing for centerf0lds before that.

BUT this does not necessarily mean that you can get the full centerf0ld image for every Pl@ymate since 2000 in this way. For example, for Tiffany Toth (Miss September 2011), a few whole columns on the right are missing, so you'd only get a truncated image once you stitch all the tiles together. Unfortunately, what seems like a simple oversight when the web page in question was put together means that only 476 tiles were linked back to the source (the rabbit company's site) when there are probably 544 tiles (my guess) for this one image. You can tell something is definitely amiss since the tiles on the right-edge are 256 pixels wide - the same as a whole tile.

It's a pity for anyone late to the party trying to get these zoom centerf0lds as close to the source as possible. There are many of these zoom CF images out there, including even a thread on this forum, but either they are not of the same quality or you cannot be sure that they have not been edited in some way by a third party (which, judging by Exif tags, most of them have been). This might not matter to most, but might bother those who collect such things.

Originally Posted by retroanalyst

The convert command from the ImageMagick suite can stitch these together, I did it the following way (this assumes a shell like bash):

for column in {0..15}; do convert -append 6-$column-{0..34}.jpg ${column}.jpg; done
convert +append {0..15}.jpg Dorothy_Stratten.jpg

This works well, but do note that the stitching process here is not lossless because the tiles are first decoded, then stitched together and finally the output is encoded as jpeg (which is a lossy format). There will be a loss in quality even if the jpeg quality is set to 100. One could use a lossless format as output but then the file size would be unreasonably large.

Ideally all the image data should just be combined without any decoding and re-encoding taking place. Some people might not care about this, but given that these are high-resolution images it could be argued that it is an important point.

Stitching the tiles together in a lossless way can be done by using recent versions of jpegtran, as explained here.

In the case of these centerf0ld tiles, they are all somewhat conveniently named (6-$C-$R.jpg or sometimes 5-$C-$R.jpg where $C and $R are the 0-based indices for the column and row respectively) that the stitching can easily be done on one (long) line in a shell like bash. Since all the tiles (except those on the right and bottom edges) are 256x256 pixels, the only other info needed in advance is the name and dimensions of the bottom-right corner tile. As an example, assuming that this tile is called 6-15-33.jpg with dimensions 120x192 (so the whole image consists of 16 * 34 = 544 tiles) and no other similar-named image files exist in the same working directory, the stitching can be done as follows (using GNU Parallel for the loop and a simple progress bar):

jpegtran -crop $((15*256+120))x$((33*256+192))+0+0 -outfile big.jpg ?-0-0.jpg && \
parallel -j1 --bar jpegtran -drop '+{=s/^\d-(\d+).*/$1/;$_*=256=}+{=s/^\d-\d+-(\d+).*/$1/;$_*=256=}' {} -outfile big.jpg big.jpg ::: ?-*-*.jpg && \
jpegtran -opt -outfile big.jpg big.jpg

This runs three commands. The first one takes the top-left tile as a base and "crops" it which effectively extends the canvas to the correct final dimensions (the numbers here need to be adjusted for each big image as explained above). The second command is a loop which losslessly inserts ("drops") each tile on to the big image at its own position (derived from the file name). This can take a while but the parallel command provides a nice progress bar. The third command simply optimises the final image so that it has a smaller file size.

(The usual disclaimer applies: you probably should not blindly run these commands without having some idea of what they do. YMMV etc.)