The Adventure Continues | osp blog

osp blog

The Adventure Continues

(A post for readers with some F/LOSS stamina)

After producing a flyer for VJ10 in Inkscape without too much trouble, we confidently decide to do a poster next. With a three week programme of lectures, workshops and installations following four interwoven themes, scattered over five locations and of course translated in three languages (Dutch, French and English), VJ10 might not be the easiest event to fit on a poster but Harrisson courageously dives in.

Experimenting with the spray paint tool he soon produces a playful image mixing and mapping the complicated Festival program. But than the trouble begins.

Finalising corrections and inevitable changes, it becomes harder and harder to navigate the document in Inkscape until we can only view it in preview mode. We also learn to split text and image into separate layers and work most of the time with the image layer switched off.

To be able to check the document, we need a full size print and this we manage with the help of poster (although our colour printer chokes on the resulting 450 MB document).

Our solution to produce PDF's with the help of the Apache Batik library is for some reason failing, so we tile the document on the basis of a poorly rendered EPS. In itself not such a problem but we know to expect problems in a later stage.

The problem with Batik originates in the 'flowroot' element that has been added in the SVG specification 1.2. Inkscape incorrectly refers to version 1.0 and Nicolas figures out that when you open the Inkscape document in a text editor and change the reference to SVG 1.0, Batik stops complaining.

But... the problem itself does not go away - the resulting PDF contains black blocks in the place of text.

We figure that 'flowroot' (text that is in a 'dynamic' textbox) is still the problem so we manually select each textbox and un-flow it. Converting the text to outline would also do the job but both solutions create a new problem. We have used colours to distinguish French, Dutch and English texts from each other but when different colors/languages placed in the same text frame are converted, they end up having only one single color.

We decide to give up on colour coding (it would mean to split all languages into separate textboxes - this is too much work and too risky with so little time left for checking and corrections), convert all text to black and mark the transitions with hearts.

Now we have a document with no 'flowroot' left, we still need to convert the SVG to PDF. With the size of this document (600 x 840 mm) Batik starts to run out of memory.

After trying to open the background image in Gimp, importing the whole .svg in Scribus, exporting the background as bitmap from Inkscape and even considering to piece screenshots of the blurs together...

Nicolas helps us out again and explains how we can assign more memory to Java so we can rasterize the file.

java -Xmx1152M -jar batik-rasterizer.jar -m application/pdf poster_vjx_verso.svg

The computers we normally work on, have not enough memory for this so we decide to descend to Constant's cellar and work on the powerful dual core machine that has been custom built by the Open Source Video team to process video files.

We have forgotten to switch on the image layer in the .svg file so we need to open the document in Inkscape first before we can pass it on to Batik. We quickly install Inkscape on this machine through the Synaptic package manager but unfortunately it defaults to version 0.44 which does not handle blur/transparency at all. The Inkscape website offers Linux Autopackages of version 0.45.1 but these fail to install on this machine (this can be the result of the specific set up of the system which is optimized for video use) and we are getting a bit desperate.

In the mean time, Nicolas has managed to convert the file on his machine and has uploaded it to a webserver. The verso of the poster is luckily converted quickly on this fast machine and we are finally ready to hand our files to the printer the next morning.

The printer experiences problems trying to output the file on high resolution, and explains us that this is caused by our 'non standard PDF'. It is hard to be sure whether that is true, or whether we have simply handed in a non-Adobe PDF but in the end he decides to open the file in Photoshop and RIP it from there... :-(

On Friday morning, we receive an e-mail from our printer. He basically tells us that there will be a two day delay (this is partially because of the unconventional folding we chose) but we are also asked to pay an additional 60 euros for having our PDF opened in Photoshop.

The proof we receive on Friday afternoon shows a reasonably well-rendered image, with one surprising detail: the size of the poster has decreased by 98%; and as a result the outer margins are too large.

Another proof will be made with the mistakes corrected and we're expected to do a last check tomorrow morning at 08:00. The adventure continues...