Input selection options

[LXConcepts]

Would be nice if the input selection in the advanced output would allow for shapes other than rectangles, or even better if the points could be manipulated

trying to map a bunch of diamond/triangular (or even worse...circular) shaped objects when combined with rectangular objects doesn't seem to work out so well...

I have several work-arounds for dealing with this kind of thing (typically multiple displays, or rotating the source content and using multiple layers), but it would be nice to be able to just cut everything up in the one window.

[Joris]

Input masking for slices is on the to-do list for Res 5.

For the time being you can perhaps work around this by masking your layers?

[LXConcepts]

like i said, there's several work arounds, but they're not very convenient

I'm not particularly looking for masking per say....I'm more looking to change the slice shape on the input selection

for example, this is my input selection window:

and this is my output slicing;

which gives me this as an over all output (using one layer to everything, for purposes of the example);

The diamond shaped bit will never map properly with the two sides given the current input selection options. This can be corrected by duplicating the layer, rotating the copy by 45 degrees and then lining it up, although because the output of the side two slices are warped on the output, it still won't line up entirely properly when a continuous image is played over all three slices, although can get close enough to look correct.

If the input selection points were adjustable to be able to mirror the output slices, this would be incredibly easy to map correctly. At the current time, it requires several layers and lots of slicing. Making a triangular piece is quite difficult when you only have rectangles to work with! Circles are even harder for a non-warped image!

Basically you have to break down your mapping target into a series of rectangles in order to get joining surfaces to line up. It's easier to fake when the items you're mapping aren't touching, as there's some leeway with the pixels that make up the spaces to fool the eye into seeing it correctly.

Masking doesn't really come into play with this sort of thing at all, it's entirely the inability to select content relative to what's actually being output. In the case of the actual object I'm mapping, it's a series of 3d structures with two sides and a diamond shaped top surface. The sides will generally display the same image, while the top may have a separate images or be part of the entire structure.

[Joris]

Thanks for the clear explanation.

I can confirm that masking and rotation for input slices is coming in the future.

To be 100% clear, note that this does not give you complete control over the output transformation points. For obvious reasons the output transformation will remain a quad.

[LXConcepts]

what about alternate input slice shapes? ie. triangle, circle, fully adjustable points/grid?

[Joris]

Masking on the input gives you all the options to create triangles, circles, hexagons, stars or any other arbitrary kind of shape. Or is that not what you are talking about?

[LXConcepts]

Masking on the input gives you all the options to create triangles, circles, hexagons, stars or any other arbitrary kind of shape. Or is that not what you are talking about?

not what im talking about...

with being able to manipulate the points on the input selection side, you're able to map without needing to know the exact pixel dimensions of what you're mapping

makes it way easier as there's less math involved

also, if you're mapping objects that are touching each other (ie. the above example), masking won't help unless you mask the entire outside of the object, in which case you run into the same problem with input slicing as you're limited to a square.

Also, you're only able to mask on the output section, not the input....I realize I can import a custom layer (or comp mask) mask, but that's incredibly limited and would require making a new mask every day, as my objects may not be the same distance apart. It's also limiting when I want to play different clips across my outputs.

ie, from the above example, playing a different clip on the diamond shaped portion while playing something else on the lower portion, isn't possible using masks

[Joris]

Sorry, you lost me.

I'm talking about the upcoming input masking feature.

I know you can't mask on the input in the current version, but if that were possible in the future, how does that still not give you the freedom you want? I'm not seeing what's so fundamentally different about having a triangular shaped slice, and taking a normal quad and drawing a triangle in there with a mask.

[LXConcepts]

quite simply, mapping a single object like this with a single clip, you can use a mask;

http://upload.wikimedia.org/wikipedia/c ... ntagon.svg

now if that shape changed sizes every night, you'd have to remake a mask every night to fit the outside of the shape- not a big deal as you just scale it up or down

now what if you wanted to divide it into five triangular pieces, like this;

playing a different clip over each section is easy - one layer per slice, route slice a to 1, b to 2, etc, use the output transforms to make 5 triangular sections and you're done. If this is the extent of what you're doing, the current config works fine.

now what about playing two clips over the sections, with one clip hitting 3 sections, and another hitting 2, both sections being mapped? ie, clip a hitting 1, 3 + 5, where clip b hits 2+4

you run into problems as your pixels no longer line up properly, as you have a square input going to a triangle output....you could do two masks (one for clip a, one for clip b), but you've now doubled your work every night

now expand outwards into more complex shapes, or more complex playback configurations...if you're mapping a single clip over everything or different clips on every section, masks will work. if you're mapping the same clip across more than 1 section, with another clip in between it anywhere, an output mask will not work. you could make a mask for every layer setup you plan on using every night, but that becomes a ton of work making new masks all the time.

for example;

if day 1 I'm working on a stage that's 40' wide x 20' high (depth doesn't particularly matter), and running an output resolution of 1920x1080, i'm looking at 48 pix by 54 pix per square foot. Day 2 my stage changes to 30' wide by 15' high, changing my p/sf to 64 pix by 72 pix. As my mapping objects are now closer together due to a smaller stage, I have to change my entire advanced output (both the input selection and output transformation) to make it line up correctly with the new configuration of objects.

The more layers I'm working with, the more masks I would need to create every day in order to do this, with a manipulative input selection (or the ability to copy output transformation into the input selection), you can eliminate the need for this kind of thing as the output and input always line up

say I have 40 slices on stage, sometimes they all play one video, sometimes there's multiple videos cut across all the slices, sometimes each slices does it's own thing.....while it's currently possible to do this in resolume, it's a major hassle as you increase your slice count as you need more and more masks- one for every layer in your show, as each slice gets a layer.

on that note....output ganging would also be nice (i want layer 1 to go to slice 2,5,6,7,8 and 12), and dmx (and osc, and midi) controllable slice selection (for cue 1 i want layer 1 to play to slice 2,4,and 5, then in cue 2 i want layer 1 to play to slice 4,7, and 8)

currently both these things have to be done by adding more layers, and essentially you're playing the same clip multiple times just to get different pieces of it to different sections of your output.

make sense?

[Joris]

Thanks again for the lengthy explanation.

I understand and acknowledge that this would be very tedious to do with Photoshop and layer masking, and impossible with just *output* masking.

I'm still not sure if you've answered my question though. If you have the ability to arbitrarily draw shape masks on your *input* slices, surely that would make this kind of setup possible and easily adaptable?