| 0,0 → 1,96 |
|---|
| |
| Implementation detail differences Daniel Marschall |
| ================================= 02 January 2018 |
| |
| FilterFoundry tries to be as compatible with FilterFactory as possible. |
| However, results are usually not 100% equal, because functions like |
| cos, sin, sqr, etc. have different accuracy due to the underlying |
| implementation. |
| |
| Furthermore, there are following known differences between Filter Foundry |
| and Filter Factory: |
| |
| i, u, v |
| ------- |
| |
| Filter Foundry <1.7 uses the same formulas as in Filter Factory: |
| |
| i=((76*r)+(150*g)+(29*b))/256 Output range is 0..254 |
| u=((19*r)+(37*g)+(56*b))/256 Output range is -55..55 |
| v=((78*r)+(65*g)+(13*b))/256 Output range is -77..77 |
| |
| Filter Foundry 1.7 uses more accurate formulas: |
| |
| i=(299*r+587*g+114*b)/1000 Output range is 0..255 |
| u=(-147407*r-289391*g+436798*b)/2000000 Output range is -55..55 |
| v=614777*r-514799*g-99978*b)/2000000 Output range is -78..78 |
| |
| Both formulas follow the same YUV standard, but have different accuracy. |
| |
| |
| get(i) (Testcase getput.afs) |
| ------ |
| |
| Filter Foundry: |
| get(x)=0 if x>255 or x<0 |
| |
| Filter Factory: |
| get(x)=x if x>255 or x<0 |
| [Note: The result "x" was most likely not intended but a result of an undefined behavior] |
| |
| |
| r, g, b at empty canvas (Testcase emptycanvas.afs) |
| ----------------------- |
| |
| In Filter Factory, an empty (transparent) canvas of a new file is initialized as r=g=b=0 |
| |
| Filter Foundry initializes it as r=g=b=255 |
| |
| |
| rnd(a,b) and rst(i) (Testcases rst_*.afs) |
| ------------------- |
| |
| Filter Foundry's implementation of rst(i) (undocumented function that sets the seed for the PRG) |
| differs from the implementation of Filter Factory. |
| |
| 1. In Filter Foundry, the function rnd(a,b) retrieves a random number at realtime; therefore, if the |
| seed is changed via rst(i), there is an immediate effect on the next call of the rnd(a,b) function. |
| For example, following filter would generate an one-colored picture without any randomness: |
| R=rst(123), rnd(0,255) |
| G=rnd(0,255) |
| B=rnd(0,255) |
| |
| In Filter Factory, the rnd(a,b) function is more complex. |
| (The analysis of Filter Factory's rnd and rst function is not yet completed, so |
| following notes might not be 100% accurate.) |
| It seems like there is a buffer with 56 random integers that will be refilled regularly. |
| The rst(i) function seems to change the seed for the next batch of random numbers, |
| so there is not immediate effect on the rnd() calls. |
| |
| 2. Furthermore, it seems like in Filter Factory, if rst(i) is called multiple times |
| with the same argument, there will be no effect. |
| So, while following filter would generate random vertical bars in Filter Foundry, |
| it would generate a picture with random pixels in Filter Factory: |
| R=(x==0) ? rst(123) : 0, rnd(0,255) |
| G=0 |
| B=0 |
| |
| In Filter Foundry, every call of rst(i) will perform an update of the seed. |
| So, if you like to have a random-pixel picture with a specific seed, |
| you must make sure that rst(i) will only be called once. |
| You would need to write your filter like this |
| R=(x==0 && y==0) ? rst(123) : 0, rnd(0,255) |
| G=0 |
| B=0 |
| if you want generate a similar result as Filter Factory outputs with the filter |
| R=rst(123), rnd(0,255) |
| G=0 |
| B=0 |
| |
| 3. In Filter Foundry, the argument i of the function rst(i) is limited to the |
| type "unsigned int" (argument of the function srand() in the C StdLib), |
| so the allowed range is 0..4294967295. |
| |
| In Filter Factory, the argument i must be between 0 and 32767, inclusively. |
| If the argument not within this range, the operation "and 0x7FFF" will be applied to it |
| to extract the low 15 bits. |