Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[jfsh]

Thanks for that note. I will have to compensate for that menu-loading initial lag somehow.

How many pages of materials do you typically see in your large, mature forts' materials list?

Rimslaughters currently has 10 pages of materials.  Others might have more, but probably not a lot more.  The material I want can frequently be on the first page in one location, and nearly on the last very close by, depending on where I am building in relation to my block and bar stockpiles.  It is also more like 5 seconds to build the materials list, which starts from the moment you pick the construction type (e.g. "floor").

[jfsh]

Just realized one problem: my graphics pack doesn't use the background highlights.  See:



Obviously I could always change graphics packs, but just an FYI.

[joelpt]

Just realized one problem: my graphics pack doesn't use the background highlights.  See:



Obviously I could always change graphics packs, but just an FYI.

Crap.

Probably time to rethink things a bit. Thanks for pointing that out.

[joelpt]

Just realized one problem: my graphics pack doesn't use the background highlights.  See:

(snip)

Obviously I could always change graphics packs, but just an FYI.

Crap.

Probably time to rethink things a bit. Thanks for pointing that out.

Well, it takes another mouse click, but I think this approach would work:

Memorization:
1. User ensures desired mat is highlighted
2. User clicks first letter of mat ('g' in 'gold bars')
3. User clicks last letter of mat ('s' in 'gold bars')
4. Take a few-pixel-tall screen clipping of the region between the two mouse clicks - call it matclip.bmp. It is a 'fingerprint' of what the un-highlighted mat item looks like.

Then we could proceed the same way as before (send + key, do image search, repeat until we find it).

[joelpt]

To also implement Draco's 'page-at-a-time' search idea we could, during memorization, additionally ask the user to UN-highlight the mat item and then re-clip the (now unhighlighted) mat item from the same screen region that we used for the highlighted clipping. Later, we could search the mats page-by-page, looking for either the highlighted OR unhighlighted version and then selecting accordingly.

But I think there's a catch here. Sometimes, during memorization, the desired mat might be all by itself on a page in the mats list -- so it would be impossible to see an unhighlighted version of the desired mat item; unhighlighting the mat would also have to switch pages.

If this is the case, then the user prompting would probably need to look something like this (after memorizing the highlighted item):

Code: [Select]

Now, UN-highlight the current material with + or - key, but keep it visible on-screen (if possible).

  Press Y if the UN-highlighted material is now visible on-screen.
  Press N if the UN-highlighted material can't be shown (i.e. it's all by itself on a mats list page).

Not sure I like this, for a couple reasons. One, it's a little confusing (at least the way it is presented here) and may be non-obvious why QF is asking for this. Two, it's more work for the user: now we have a highlighting step, 2 mouse clicks, an UNhighlighting step, and a Y or N keypress.

Considering that the desired mat for a construction will most often appear within the first 2 pages of the mats list, I'm tending towards just implementing the more naive 'item-at-a-time' approach for now, and thus minimizing the required user interactions.


I've thought about doing memorization in such a way that a previous memorization can be reused on subsequent blueprints/Alt+D's (remembering the chosen mat for reuse). DFWall took a similar approach, requiring you to 'memorize' mats before starting the designation routine. While this would be nice for doing a lot of smaller blueprints with the same mat, it won't work if the user changes their UI zoom level, because those remembered matclip.bmp's would no longer match anything on-screen. So I think at least initially the user will just be required to freshly memorize the desired mat(s) every time Alt+D is used.

[Draco18s]

I remember that DFWall had to memorize the spritesheet used by the player before it could attempt to look for selected materials.

In any case, if you want to go with 1-click solutions, simply:

"Click highlighted material.  If no background highlight, click ON the letter."

Then just run as normal.  Find the top edge of the indicated color, if it's a letter, this will be slightly below the desired region.  Go up one pixel, note color.  Scan left and right to find where this color ends.  From click-location, scan down to edge of color, extend 1 pixel, note height.  Make clip from UL to UR+Height.

In the case of clicking the dot in the i, you'll end up with a very skinny scan region, but 99% of the time you'll still only find one match (if you don't: it's the user's fault for being dumb).

Edit:

Crap, forgot about the numbers on the right.  One moment.

Bugger.  I made a few mistakes in my assumptions.  This might not be as easy as I thought (damn anti-aliasing).

Hmm.  You could ask the user to drag-select the desired material, and use that as your clip region.

[joelpt]

I remember that DFWall had to memorize the spritesheet used by the player before it could attempt to look for selected materials.

Yeah that's what I was hoping to avoid. Also, you can zoom the whole DF GUI in/out now, which breaks image searches versus images such as the spritesheet which aren't at that exact same zoom level.

In any case, if you want to go with 1-click solutions, simply:
"Click highlighted material.  If no background highlight, click ON the letter."
(snip)

Yeah that's the approach I started considering too, but it gets very hard to determine where the edge-boundaries really are. I started thinking about things like having the user calibrate where the menu is by clicking on the "Item" header and "v: View Item" text in the menu, and then ... well, then it started getting complicated.

Hmm.  You could ask the user to drag-select the desired material, and use that as your clip region.

This is basically what I've ended up with in my latest prototype, though it's done as two mouse clicks rather than a drag operation. It then screenclips the rectangle formed by the two click-points.

I am currently coercing the clip to be at least 3px tall so there's enough to uniquely distinguish the screenclipped 'fingerprint' from any other mat-item we might encounter. I also instruct the user to "Click center of FIRST letter" and "Click center of LAST letter" to help avoid accidentally capturing an adjacent mat-item in the screenclip.

Here's an example of such a 'fingerprint':



You can probably figure out what this mat was just from looking at this clip. For QF, this will *hopefully* be 100% unique amongst a list of 100s of mats. Users can actually click closer to the top-left and lower-right corners of the highlighted mat selection to get a taller screenclip, which should be even more distinguishable should there be problems with mis-identification. I doubt this is going to be a problem though -- that fingerprint up there seems like it will be pretty darn distinct.

[joelpt]

More "fingerprint" examples from latest prototype:

[danielout]

I'm an die-hard quickfort user. I cannot express how much time it has saved me. A friend of mine wasn't as convinced. So I tried to think of a way to entice him; this is the result:
http://youtu.be/czrCvZqIMEc

ed note:  I realize it is far longer than it needs to be; I don't imagine many people will watch until the end. Also, for some reason youtube really likes to screw up the video for the first 15 seconds. Fine on my local machine, but no amount of reuploads seems to fix it after youtube does its conversion magic.

Nothing super helpful, but I figured it illustrated the utility of QF and gave a great example of why it is worth the download while being (slightly? maybe?) funny.

[joelpt]

I'm an die-hard quickfort user. I cannot express how much time it has saved me. A friend of mine wasn't as convinced. So I tried to think of a way to entice him; this is the result:
http://youtu.be/czrCvZqIMEc

ed note:  I realize it is far longer than it needs to be; I don't imagine many people will watch until the end.

Haha nice one. Especially the "reaction" webcam in the corner

Your DF macro playback looks like it's going quite slow: you may need to set [MACRO_MS:0] in your DF data/init/init.txt.

I should probably just code a check for this init setting into QF; I bet it's catching up more than a few QF2 users.

[joelpt]

The basics of the manual material selection feature are now in.

Here's a short video showing it in action. The video has no audio, but I am currently using an audio cue (playing a .wav file) when user input is required for the material memorization phase. Just need to find the right non-intrusive "attention please" sound effect

The cell syntax is "cmd:label", e.g. Cf:1, Cw:outer, ...

For instance, a constructed checkerboard floor can now be done with something like:

Spoiler: checkerboard pattern (click to show/hide)

Code: [Select]

#build
Cf:1,Cf:2
Cf:2,Cf:1


Alt+T -> build Cf:1,Cf:2#Cf:2,Cf:1
Alt+R -> 10e 10s

This feature will probably appear in QF 2.02, though likely marked as "experimental". I imagine it will take a few releases before all the kinks are ironed out (particularly with respect to non wall/floor constructions).

[joelpt]

Anybody got a mature 0.31.25 fort save file they'd like to send me? I need a midsize fort I can conduct experiments in, but I'm too lazy to actually PLAY DF long enough to get one

I downloaded Rimslaughters but just unpausing it makes my PC cry uncle.

[wrajjt]

Hey all. I just downloaded this excellent utility app in hopes of finding something that would make the creation of my borehole fortress easier (inspired by http://www.bay12forums.com/smf/index.php?topic=59656.0) but I did not find a finished blueprint, and I have no ways of creating one myself. What I need is a 41 units wide circle of up-ramps, and the "ramp-area" needs to be two units wide. Could someone put one together for me, or point me towards a way of doing so when you've got nothing that can open or create excel spreadsheets?

[Aklyon]

D'you have OpenOffice/LibreOffice? They work just as well for saving .csv, and usually as well with .xls as well.

[wrajjt]

None of those, and googledocs dont seem to be able to create an xml with 41 columns :/