From 089475041fe26964d72cb2ebc3559a36ba89a2f2 Mon Sep 17 00:00:00 2001 From: "N.N." Date: Tue, 8 Jul 2008 05:56:10 +0000 Subject: trying to import gridflow 0.9.4 svn path=/trunk/; revision=10148 --- externals/gridflow/doc/reference.xml | 1329 ++++++++++++++++++++++++++++++++++ 1 file changed, 1329 insertions(+) create mode 100644 externals/gridflow/doc/reference.xml (limited to 'externals/gridflow/doc/reference.xml') diff --git a/externals/gridflow/doc/reference.xml b/externals/gridflow/doc/reference.xml new file mode 100644 index 00000000..339862d4 --- /dev/null +++ b/externals/gridflow/doc/reference.xml @@ -0,0 +1,1329 @@ + + + + + + +
+ + +

this object is the opposite of #import.

+ + this object is not configurable because there isn't + anything that could possibly be configured here. + + + + transforms this grid into a sequence of integer messages. + + + + + elements of the incoming grid. + + + + + +

this object is another opposite of [#import], which puts + all of its values in a list.

+ + + + transforms this grid into a single message containing + a list of integers. + + + + + elements of the incoming grid. + + + + + +

this object is another opposite of #import, which constructs a symbol + from its input. The values are expected to be valid ASCII codes, but no check + will be performed for that, and additionally, no check will be made that the generated + symbol only contains characters that can be put in a symbol.

+ + + + transforms this grid into a single message containing + a list of integers. + + + generated symbol + + + +

Similar to [#join], but takes individual integers, and builds a Dim(N) vector out of it. +

+ + + The value "any" (and the only available value for now) causes an output + to produced when an integer is received thru any inlet, contrary to most + other object classes, that only act upon reception of a value thru inlet 0. + + + + how many inlets the object should have. + + + + combination of inputs given in all inlets. + this is produced according to the value of the trigger attribute. + + + + +

Triple slider for the selection of RGB values.

+ + + + + + + changes all three values (R,G,B). The grid must + be a Dim(3). + + + sends the rest of the message to each of the three sliders. + this relies on the fact that [#color] is implemented using + three [hsl] and this might not still work in the far future. + + + Produces a Dim(3) grid of RGB values. + + + + + + + how many outlets the object should have. + (depending on the version of the software, the number of visible outlets + may have been frozen to 4. If it is so, then the value of this argument + must not exceed 4; and if it is below 4, then don't use the extraneous outlets.) + + + + the input vector is split in N parts containing one number each. + numbers are sent left-to-right, that is, outlet 0 is triggered first, then outlet 1, etc. + + + + + + + + + will compute the centroid of the given grid, which + is a weighted average, namely, the average position weighted + by the pixel values. + + + + result + + + + + +

when given scalar bounds, works like a regular [for] object plugged + to a [#import] tuned for a Dim(size) where size is the number of values + produced by a bang to that [for].

+ +

when given vector bounds, will work like any number of [for] objects + producing all possible combinations of their values in the proper order. + This replaces the old [#identity_transform] object.

+ + + + + + + + replaces the "from" value and produces output. + + + replaces the "to" value. + + + replaces the "step" value. + + + + where size is floor((to-from+1)/step) + [for scalar bounds] + + + where *size is floor((to-from+1)/step) + [for vector bounds] + + + +
+ +
+ + + + + + +

This object outputs a grid by computing "in parallel" a same + operation on each left-hand element with its corresponding right-hand + element. +

+ + + + + + + + + on each element of this grid, perform the operation + together with the corresponding element of inlet 1. + in the table of operators (at the top of this document) + elements of inlet 0 are called "A" and elements of inlet 1 + are called "B". the resulting grid is the same size as the + one in inlet 0. + + + + + any grid, preferably shaped like the one that will be put + in the left inlet, or like a subpart of it (anyway the contents + will be redim'ed on-the-fly to fit the grid of inlet-0, + but the stored grid will not be modified itself) + + + stores a single int in the right inlet; the same int will + be applied in all computations; this is like sending a + Dim(1) or Dim() grid with that number in it. + + + + + + + + + + +

this object computes the square of complex numbers. + If seeing imaginary as Y and real as X, then this operation squares + the distance of a point from origin and doubles the angle between it + and the +X half-axis clockwise. (fun, eh?) +

+

used on an indexmap, this makes each thing appear twice, + each apparition spanning half of the original angle.

+ + + + + + + + + +

+

  • [#fold +] computes totals
    • +
  • [#fold inv+] is an alternated sum (+/-)
    • +
  • [#fold * 1] can compute the size of a grid using its dimension list
    • +
  • [#fold & 1] can mean "for all"
    • +
  • [#fold | 0] can mean "there exists (at least one)"
    • +
  • [#fold ^ 0] can mean "there exists an odd number of..."
    • +
  • [#fold ^ 1] can mean "there exists an even number of...".
    • +

    + + + + + + + + + + replaces every Dim(last) subgrid by the result of a cascade on that subgrid. + Doing that + with seed value 0 and operation + on grid "2 3 5 7" will compute + ((((0+2)+3)+5)+7) find the total "17". + produces a Dim(dims) grid. + + + + +     + + +

    [#scan +] computes subtotals; this can be used, for example, + to convert a regular probability distribution into a cumulative one. + (or in general, discrete integration) +

    + + + + + + + + + + replaces every Dim(last) subgrid by all the results of + cascading the operator on that subgrid, + producing a Dim(dims,last) grid. + + For example, with base value 0 and operation + on grid "2 3 5 + 7" will compute 0+2=2, 2+3=5, 5+5=10, 10+7=17, and give the + subtotals "2 5 10 17". + + + + + + + +     + + + + + + the operator must be picked from the table of two-input operators. + the grid is optional and corresponds to inlet 1. + + + + + produces a grid of size Dim(anyA..., anyB...), where numbers + are the results of the operation on every element of A and + every element of B. the resulting array can be very big. Don't + try this on two pictures (the result will have 6 dimensions) + + + + stores the specified grid, to be used when inlet 0 is activated. + + + + +

    When given a grid of Dim(3) and a grid of Dim(5) [#outer] will + produce a grid of Dim(3,5) with the selected two-input operation + applied on each of the possible pairs combinations between numbers + from the left grid and the ones from the right. for example : + (10,20,30) [#outer +] (1,2,3) will give : + ((11,12,13),(21,22,23),(31,32,33))

    + +     + + +

    think of this one as a special combination of [#outer], [#] and + [#fold]. this is one of the most complex operations. It is very useful + for performing linear transforms like rotations, scalings, shearings, + and some kinds of color remappings. A linear transform is done by + something called matrix multiplication, which happens to be [#inner * + + 0]. [#inner] also does dot product and other funny operations.

    + + + + + + + + Splits the Dim(anyA...,lastA) left-hand grid into Dim(anyA...) + pieces of Dim(lastA) size. + + Splits the Dim(firstB,anyB...) right-hand grid into + Dim(anyB...) pieces of Dim(firstB) size. + + On every piece pair, does [#] using the specified + op_para operation, followed by a [#fold] using + the specified op_fold operator and base value. + + creates a Dim(anyA...,anyB...) grid by assembling all + the results together. + + (note: lastA must be equal to firstB.) + + + the operation that combines the values from the two grids together. + this defaults to "*" (as in the matrix product) + + + the operation that combines the result of the "op" operations together. + this defaults to "+" (as in the matrix product) + + + + + changes the base value to that. + + + + + changes the right-hand side grid to that. + + + + +     + + + + + Which_dim is the number of the dimension by which the join will + occur. For N-dimensional grids, the dimensions are numbered from 0 + to N-1. In addition, negative numbers from -N to -1 may be used, to + which N will be added. + + + + The left grid and right grid must have the same number + of elements in all dimensions except the one specified. + The result will have the same number of elements in all + dimensions except the one specified, which will be the + sum of the two corresponding one. + +

    For example, joining a RGB picture Dim[y,x,3] and a + greyscale picture Dim[y,x,1] on dimension 2 (or -1) could + make a RGBA picture Dim[y,x,4] in which the greyscale image + becomes the opacity channel. +

    +     +     + + + +     + + + + any grid + + + a bang is emitted every time a grid transmission ends. + + + + + + + + + any grid + + + a grid of the same shape containing all the same + values after type conversion. note that while casting to + a smaller type, values that are overflowing will be truncated. + + + + + + any grid + + like [#redim] but always produce a 1-D grid + with the same total number of elements. + + + + + any grid + +

    splits a Dim[A...,B] grid into Dim[B] vectors, + producing new Dim[B] vectors that each contain numbers from + 0 to B-1 indicating the ordering of the values. The result is + a Dim[A...,B] grid.

    +

    for example, connecting a [#grade] to a [#outer ignore {0}] + to a [#store] object, storing a single vector into [#store], and + sending the same vector to [#grade], will sort the values of the + vector. however for higher-dimensional grids, what should go + between [#store] and [#grade] to achieve the same result would + be more complex.

    +

    you may achieve different kinds of sorting by applying various + filters before [#grade]. the possibilities are unlimited.

    +

    if you plug [#grade] directly into another [#grade], you will + get the inverse arrangement, which allows to take the sorted values + and make them unsorted in the original way. note that this is really + not the same as just listing the values backwards.

    +     +     + + + + + + any grid + +

    transforms a Dim[A...,B] grid into a Dim[A...,B-1] grid. + There is a projection plane perpendicular to the last axis and + whose position is given by the "depth" parameter. Each vector's + length is adjusted so that it lies onto that plane. Then the + last dimension of each vector is dropped.

    + +

    useful for converting from 3-D geometry to 2-D geometry. Also + useful for converting homogeneous 3-D into regular 3-D, as + homogeneous 3-D is really just regular 4-D...(!)

    +     +     + + + + + + + + swaps the two specified dimensions; dimension numbers are as in [#join]. + + + + + + + + + produces on outlet 0 a linear recurrent fading according to the flow of + incoming messages. For example, if rate=5, then 20% (one fifth) + of each new message will be blended with 80% of the previous output. + + + + + + + + + + produces on outlet 0 a piecewise-linear nonrecurrent fading according to the flow of + incoming messages. For example, if maxraise=2 and maxdrop=4, then with each + new message an output is produced that is at most 2 more or 4 less than the + previous output. + + + + + + + Whichdim is the number of the dimension by which the reverse will + occur. For N-dimensional grids, the dimensions are numbered from 0 + to N-1. In addition, negative numbers from -N to -1 may be used, to + which N will be added. + + +
    + +
    + + + + a list specifying a grid shape that the numbers + will fit into. + (same as with [#import]) + + + + the elements of this grid are serialized. if the resulting grid + must be larger, the sequence is repeated as much as necessary. + if the resulting grid must be smaller, the sequence is truncated. + then the elements are deserialized to form the resulting grid. + + + + + this grid is a dimension list that replaces the one + specified in the constructor. + (same as with [#import]) + + + + + redimensioned grid potentially containing repeating data. + + + +

    example: with a 240 320 RGB image, [#redim 120 640 3] will visually + separate the even lines (left) from the odd lines (right). contrary + to this, [#redim 640 120 3] will split every line and put its left half + on a even line and the right half on the following odd line. [#redim] + 480 320 3 will repeat the input image twice in the output image. + [#redim] 240 50 3 will only keep the 50 top lines.

    + +     + +

    A [#store] object can store exactly one grid, using the right + inlet. You fetch it back, or selected subparts thereof, using the left + inlet.

    + + + + + + + + the stored grid is fully sent to the outlet. + + + in this grid, the last dimension refers to subparts of + the stored grid. sending a Dim(200,200,2) on a [#store] + that holds a Dim(240,320,3) will cause the [#store] to handle + the incoming grid as a Dim(200,200) of Dim(2)'s, where each + Dim(2) represents a position in a Dim(240,320) of Dim(3)'s. + therefore the resulting grid will be a Dim(200,200) of + Dim(3) which is a Dim(200,200,3). in practice this example + would be used for generating a 200*200 RGB picture from a + 200*200 XY map and a 240*320 RGB picture. this object can + be logically used in the same way for many purposes + including color palettes, tables of probabilities, tables + of statistics, whole animations, etc. + + + + + replace the whole grid, or a subpart of it (see other options on inlet 1) + + + + (Future Use): + makes it so that sending a grid to inlet 1 detaches the old buffer from [#store] + and attaches a new one instead. This is the default. + + (Future Use): + makes it so that sending a grid to inlet 1 writes into the existing buffer of [#store]. +

    + example: suppose you have [#store {10 240 320 3}]. then "put_at 3" + will allow to write a Dim[240,320,3] grid in indices (3,y,x,c) where y,x,c are indices of the incoming grid; + in other words, if that's a buffer of 10 RGB frames, you'd be replacing frame #3. Furthermore, + it also allows you to write a Dim[n,240,320,3] grid at (3+f,y,x,c) where f,y,x,c are indices of the incoming grid, + replacing frame #3, #4, ... up to #3+n-1. Here n is at most 7 because the last frame in the buffer is #9. +

    +

    that way of working extends to other kinds of data you'd put in Grids, in any numbers of dimensions; + because, as usual, [#store] wouldn't know the difference. +

    +     +     + + grids as stored, as indexed, or as assembled from multiple + indexings. + +     + + + + {height width} pair. + + + + a 3-channel picture to be scaled. + + + a {height width} pair. + + + a scaled 3-channel picture. + + + + + + + factor is optional (default is 2). + if it's a single value, then that factor is to be used + for both rows and columns. + + + + duplicates each pixel several times in width and several times in height, + where the number of times is determined by the factor described above. + twice those of the incoming grid. It is several times faster. + + + sets factor + + + + + + + + + + + + factor is optional (default is 2). + if it's a single value, then that factor is to be used + for both rows and columns. + + + + Scales down picture by specified amount. (See scale factor above) + + + + sets scale factor + + + + + + + + +

    typically you plug a [#for] into this object, + and you plug this object into the left side of a [#store]. it will + scatter pixels around, giving an "unpolished glass" effect.

    + +

    if you put a picture in it, however, it will add noise. The + resulting values may be out of range, so you may need to clip them + using min/max.

    + + + same as inlet 1 + + + + a coordinate map. + + + a spread factor. + + + a coordinate map. + + + +

    [#spread] scatters the pixels in an image. Not all original pixels + will appear, and some may get duplicated (triplicated, etc) + randomly. Some wrap-around effect will occur close to the edges. +

    + +

    Sending an integer to inlet 1 sets the amount of spreading in + maximum number of pixels + 1. even values translate the whole image + by half a pixel due to rounding.

    + +     + + +

    performs rotations on indexmaps and polygons and such.

    + + + + + + + rotation angle; 0...36000 + + + + +     + + +

    if you chain indexmap (coordinate) transformations from outlet 1 + to inlet 1, then sending an image in inlet 0 will emit its + deformation out of outlet 0.

    + + + + + +     +
    + +
    + +

    Returns list of dimensions as a grid. Given a grid sized like Dim(240,320,4), + [#dim] will return a grid like Dim(3), whose values are 240, 320, 4.

    + + + no arguments. + + + + ignores any data contained within. + sends a grid dim(length of dims) containing dims. + + + + + the list of dimensions of the incoming grid. + + +     + + +

    gives a symbol representing the numeric type of the grid received. +

    + +     + + + GUI object equivalent to [print] and [#print]. + + Displays the received message in the box, resizing the box so that the message fits exactly. + + +
    + +
    + +

    This object is useful for color correction. For each pixel + it takes it apart, looks up each part separately in the colormap, + and constructs a new pixel from that. You may also color-correct + colormaps themselves.

    + +

    Only works for things that have 3 channels.

    + +

    Note: if you just need to apply a palette on an indexed-color + picture, you don't need this. Just use #store instead.

    + + + + picture + + + + + colormap ("palette") + + + + picture + +     + + + + + + + + + + + + + + + + + + + + +

    note: may change slightly to adapt to actual video standards.

    + + + + + + +     + + +

    note: may change slightly to adapt to actual video standards.

    + + + + + + +     +
    + +
    + +

    this is the object for blurring, sharpening, finding edges, + embossing, cellular automata, and many other uses.

    + + + + + + + + + + + splits the incoming grid into dim(rest...) parts. + for each of those parts at (y,x), a rectangle of such + parts, centered around (y,x), is combined with the + convolution grid like a [#] of operation op_para. Then + each such result is folded like [#fold] of operation + op_fold and specified base. the results are assembled + into a grid that is sent to the outlet. near the borders of + the grid, coordinates wrap around. this means the whole grid + has to be received before production of the next grid + starts. + + + + + + this is the convolution grid and it gets stored in + the object. if rows2 and/or columns2 are odd numbers, + then the centre of convolution is the middle of the convolution + grid. if they are even numbers, then the chosen centre will + be slightly more to the left and/or to the top, because the + actual middle is between cells of the grid. + + + + + + + +     + + + + same as inlet 1. + same as inlet 2. + + + + + + produces a grid like the incoming grid but with + different constrast. + +

    [#contrast] adjusts the intensity in an image. + resulting values outside 0-255 are automatically clipped.

    +     + + + this is the secondary contrast (inverse whiteness). + it makes the incoming black + correspond to a certain fraction between output black and the + master contrast value. no effect is 256. default value is 256. + + + + + this is the master contrast. it makes the incoming white + correspond to a certain fraction between output black and output + white. no effect is 256. default value is 256. + + + + + + + +     + + +

    [#posterize] reduces the number of possible intensities in an image; + it rounds the color values.The effect is mostly apparent with a low + number of levels.

    + + + same as inlet 1 + + + + + + produces a posterized picture from the input picture. + + + + + + this is the number of possible levels per channel. the + levels are equally spaced, with the lowest at 0 and the + highest at 255. the minimum number of levels is 2, and the + default value is 2. + + + + + + +

    example: simulate the 216-color "web" palette using 6 levels. + simulate a 15-bit display using 32 levels.

    +     + + +

    makes medium intensities brightest; formerly brightest colours + become darkest; formerly darkest stays darkest. This filter is linear: + it's like a 200% contrast except that overflows are mirrored + instead of clipped or wrapped.

    + + + + + + + + +     + + + + + + result from a [#for {0 0} {height width} {1 1}] + + + + + + checkered pattern of 50%/75% greys + in 8x8 squares + + + + + + + + + a picture that has an opacity channel. + will be used as foreground. + + + + + + a picture that has NO opacity channel. + will be used as background. + + + + + + a picture that has NO opacity channel. + the opacity channel of the foreground is used as + a weighting of how much of either picture is seen + in the result. + + + + + + + + Normally you would use the "put" operator here; + but abnormally I recommend + and ^ for psychedelic effects. + + + + + + + picture onto which another picture will be superimposed. + + + if enabled, inlet 1 picture will be repeated to cover the inlet 0 picture. + + + if enabled, inlet 1 picture will be combined with inlet 0 picture using + the selected operator, + and then blended with inlet 0 picture according to transparency of + the inlet 1 picture, and then inserted in the result. + if disabled, the blending doesn't occur, as the transparency level + is considered to be "opaque". note that with alpha enabled, + the last channel of inlet 1 picture is considered to represent transparency. + + + + + picture that will be superimposed onto another picture. + + + + + position of the inlet 0 picture corresponding to top-left corner + of inlet 1 picture. + + + + + resulting picture. + + + + + + + + Normally you would use the "put" operator here; + but abnormally I recommend + and ^ for psychedelic effects. + + + + + + + picture on which the polygon will be superimposed. + + + + + color of each pixel + + + + + vertices of the polygon. + + + + + + modified picture. + note: starting with 0.7.2, drawing a 1-by-1 + square really generates a 1-by-1 square, and + so on. This is because the right-hand border of a + polygon is excluded, whereas it was included + before, leading to slightly-wider-than-expected polygons. + + + + + +

    inlet 2 receives a font grid, for example, [#in grid file lucida-typewriter-12.grid.gz]

    +

    inlet 1 receives a 2 by 3 matrix representing the colours to use (e.g. (2 3 # 0 170 0 255 255 0) means yellow on green)

    +

    inlet 0 receives a bang, transforming the data into an image suitable for #draw_image.

    +     + +

    inlet 1 receives an angle (0..36000)

    +

    inlet 0 receives a RGB picture that gets hueshifted by a rotation in the colorwheel by the specified angle; it gets sent to outlet 0.

    +     +
    + +
    + + Transforms linear counting (0, 1, 2, 3, 4, ...) into a back-and-forth counting (0, 1, 2, 1, 0, ...) + from 0 to a specified upper bound. + + + + + + + + + a value to be transformed. + If, for example, top=10, then values 0 thru 10 are left unchanged, + values 11 thru 19 are mapped to 9 thru 1 respectively, and 20 thru 30 + are mapped to 0 thru 10, and so on. + + + + + + + which clock to use. "real" uses wallclock time. "user" uses + the amount of time spent in the process. "system" uses the + amount of time spent in the kernel on behalf of the process. + "cpu" uses the Pentium clock, which is like a more precise + version of "real" if you have a Pentium. + + optional + + + + + + Times at which bangs are received are stored until a large + enough sample of those is accumulated. Large enough is defined + to be whenever the timespan exceeds one second. Then a report + is made through the outlet. + + + messages other than bangs are ignored. + + + + + non-detailed mode only. + this is the messages-per-second rating. + + + detailed mode only. + this is: messages-per-second, followed by five values of + milliseconds-per-message: minimum, median, maximum, average, + standard deviation. + (the average happens to be simply 1000 divided by the + messages-per-second, but it is convenient to have it anyway) + + + + +

    + This object returns the Unix timestamp. The first + outlet does so with ASCII, the second in seconds and the third outlet + outputs the fractions of seconds up to 1/100 000 th of a second which is useful for creating + filenames. +

    + + Outputs the time and date in ASCII format + Outputs the Unix timestamp in seconds + Outputs the fractions of a second up to 10 microseconds (?) (actual precision is platform-dependent afaik) +     + +

    + This object produces HPGL instructions in ASCII form + that can be sent to the comport object in order to control an HPGL + compatible plotter. +

    + + Outputs the HPGL commands in ASCII format +     +
    + +
    +

    those classes emulate jMax functionality, for use within PureData.

    + + +

    Every incoming message is sent to inlet 1 and then sent to + inlet 0 as well. Messages remain completely unaltered. Contrast + with PureData's "t a a" objects, which have the same purpose but + transform bangs into zeros and such.

    + + + + +     + + + Outputs N messages, one per list element, in order. + + + + + + + + + outputs the number of elements in the incoming list. + + + + + + + + Outputs one element of the list, as selected by "index". + Also accepts negative indices (e.g.: -1 means "last"), unlike jMax. + + + + + + + + + + Outputs consecutive elements of the list, as selected by "index" and "length". + Also accepts negative indices (e.g.: -1 means "last"), but unlike jMax. + + + + + + + + + + Outputs the stored list followed by the incoming list, all in one message. + + + + + + + + + Outputs the incoming list followed by the stored list, all in one message. + + + + + + Outputs the incoming list, from last element to first element. + + + + Like [spigot], but turns itself off after each message, so you have to turn it on + again to making it pass another message. + + + + + + + outputs b-a + + + + + + + + + outputs b/a + + + + + (This is not in jMax, but is there to help port $* messageboxes) + + + Like [listprepend], but operates on whole messages, that is, including the selector. + + + + + (This is not in jMax, but is there to help port $* messageboxes) + + + Like [listappend], but operates on whole messages, that is, including the selector. + + + + + + Compatible with jMax's [demux]. + + number of outlets + initial selected outlet + + + Routes a message to the active outlet. + + + Selects which outlet is active. + + + + please use shunt instead (name conflict with another Pd external) + + + + + + + a value to be sent to one of the outlets. The first outlet is for values + smaller than the first argument; else the second outlet is for values smaller + than the second argument; and so on; and the last outlet is for values greater + or equal to the last argument. + + + sets the corresponding separator in the separator list. + + +
    +     -- cgit v1.2.1