Question about importing Devonthink images

I recently dragged an image from DevonThink to Tinderbox – and was amazed was how smooth it was.

I saw a preview, a note, attributes filled out (including a link to the original DT item!) … just great!

But I immediately had a follow-on curiosity: what exactly happened here – is TB displaying a preview of a “external” image, or is this “internal”, i.e. within the TB file?


If you see an image in the $Text of the note the image has been ingested into the TBX document’s data. If the image is only seen in Preview mode then the information is being drawn from elsewhere via whatever export template you have set for that note.

The problem diagnosing this is you are describing the visual output but which could arise from a number of different routes. We know you see info in the Text ‘Preview’ pane and that the data originated in /03/2014 but it is important to understand this tells your fellow users very little diagnostically. I say that not to be harsh but as an explanation that above it would have been more useful to, for instance, shown a screen drab of the text pane , i.e the Displayed Attributes and the text.

Bear in mind that the preview pane in Tinderbox is essentially an internal web browser. So you may think you are working in RTF, XML, Markdown, YAML, etc, but what you are seeing in preview is rendered HTML.

Want to know where the image comes from, check the HTML tab which is the HTML extraction of whatever your note +template have created. For instance if the src attribute of the `<img.> (image) tag, i.e. the code element causing the image in the rendered HTML, has a path pointing to DEVONthink then likely the image is being supplied by DEVONthink. If there is no path and just an image name likely the image is embedded in $Text (which you can easily verify by looking at text.

Preview pane is fine when all works. When the result looks wrong stop looking at Preview as it only indicates a problem but is near useless at further diagnostics. At this point look at the source code in the HTML tab. Fix that and you preview will be back on track. Don’t understand the HTML code? Then that and not the preview is the thing to share with fellow users. The code gives us clues. The preview gives us next to nothing with which to help you.

The above lesson is even truer if doing complex exploits as shown in some of the recent video lessons. Preview is a preview and not a diagnostic tool! The ‘HTML’ pane is your initial diagnostic tool.

Confusion arises because the preview was originally what would be exported, but increasingly is being used as an alternate view type. But, even if the preview is used for data never exported, the data rendered in the preview is HTML.

I’ve also noticed people getting confused by trying to export by copy/pasting preview content. Let’s say you are copying into Word (horrid tool, but an all too common choice):

  • Copy/paste from the preview pane. You are pasting an RTF render of the data. IOW, all semantic structure like heading markers is missing.
  • Copy /paste from the HTML pane. You paste HTML, with CSS if any, which Word then translates into RTF but using the structure.

Thus in the first case the second level headings, for example, may all look the same in WYSIWG render but are styled per instance. In the second case, if you’ve use semantic structure in your HTML (e.g. a styled <h2> element) then all H@ headings should use the same style and my thus also all be re-styled by editing the style. In the first case, every second-level heading would need to be individually edited to change the style. Garbage in—garbage out. The app (in this case either Tinderbox or Word) can’t guess what we can’t be bothered to tell it.

Bottom line, learn to understand the HTML pane, it will save you a lot of time if using formatted export/preview. :slight_smile:


BTW, if it is of interest, here is a write-up I’ve done that shows how to reference an image from Devonthink. You can leave it in Devonthink but view it in TBX/


Wonderful, thanks for the detailed write up, I hadn’t thought of it as “an html pane”, helps a lot.

Also, thanks for the pointer to referencing the image, I’ll usually want to do that (I think, IIUC TB shouldn’t have an arbitrary amount of large images within it, right?)

While you can have images locally stored in TBX it does create unnecessary bloat. I find the referencing approach more effective and efficient.