adding pixel itensities, clamping at max subtracting intensities, clamping at 0 taking the minimum of the two intensities inverts the pixel value doesn't seem to be used. there is an invert.c that has the same function and is the one that is called e.g. by image/invert menu takes a source, a destination and a color. Where the diff between the color and a source pixel is greater than a threshold, the destination is set to white, where it is smaller, it is set to black. Also allows to use antialiasing, but I don't understand the calculations by_color_select.c: by_color_select_color, fuzzy_select.c: seed_fill extract the data of one channel that is given as parameter and copies it to the destination row colours a row in a given colour: sets all pixels in the row to the intensities of that colour blends two rows with a blending factor >0.0 and <1.0: row1*blend + row2*(1-blend) basically same as blend = blend one row with a colour instead of blending two rows could get rid of this one, since same as blend, pixel_row or colour as second operand is the same extracts the alpha data, mixing it with mask data if there is, otherwise assuming opaque mask mix two rows by always taking the darker pixel of the two (same for alpha) inverse of darken_row takes two rows, converts to HSV and forms destination row by taking either H,S or V (depending on a mode parameter) from second row and rest from first row. Then converts back converts two rows to HLS, takes L from first row and HS from second to form result row, has a mode parameter that is ignored. Converts result back to RGB mode parameter is ignored mults two rows, and divides by max to clamp. takes minimum of the two alpha 'screens' two rows (as in the layer mode 'screen'): inverts both, multiplies and inverts the results as in the layer mode not entirely sure this is correctly implemented. was a bit uncertain about where to clamp adds values, clamping at max does exactly the same as x_add_row subtracting values, clamping at 0 does exactly the same as x_sub_row calculates the absolute difference, takes the minimum of the two alpha as in the layer mode. sets alpha to transparent (0) if it is greater than some random value, otherwise maintains it don't understand the calculations. seems to blend the two rows in a way? !!POSSIBLY WRONG!! didn't understand it, left it for bfp as for u16 as the name says multiplies intensities by scale factor doesn't seem to be used by anyone (probably replaced by scale_row_* funcs in paint_funcs_area. Also, I think, it has a bug for u8 and u16 (uses INT_MULT straight with float values, without pre-multiply) adds alpha channel to an image and makes it opaque flattens a row, i.e. blends/combines two rows using their alpha information multiplies the alpha channel onto RGB (pre-multiplies) doesn't seem to be used by anyone the inverse of multiply_alpha_row_bfp (divides alpha out of pre-multiplied values) doesn't seem to be used by anyone as name says is this ever ??use?? (also copy_gray_to_inten_a) combines mask and alpha channel storing result in the latter (mask * alpha) alpha + (1 - alpha) * mask takes gray image and copies to new (possibly expanding RGB) + adding alpha opaque the initial functions calculated the display intensities for first layer when composing an image used to display as grayscale when only one channel + no layers visible 100% same code as copy_gray_to_inten_a_row_bfp use indexed in bfp? use indexed in bfp? the combine functions are used to gradually combine the layers in the rendering process. There is a function for every possible combination of alpha channel/no alpha channel. the second source row is the one that is composed on top of the first use indexed in bfp? use indexed in bfp? use indexed in bfp? use indexed in bfp? combines RGB+RGB or gray+gray not sure what this does indexed used for bfp? not sure what this does indexed used for bfp? not sure what this does indexed used for bfp? extracts pixels from a row, pixels to extract are given by a mask. parameter cut is boolean (TRUE: cut out pixels from source, FALSE, just copy). If we cut and the source has alpha, it becomes transparent where not masked or stays the same where masked.if no alpha exists, the cut is realised by displaying the background colours where masked gimage_mask.c:gimage_mask_extract, global_edit.c:edit_copy + edit_cut indexed used for bfp? a group of copy functions from rgb/gray/indexed bfp to every other format all the functions below just wrap up the type-dependent row_functions in paint_funcs_row_*.c. They select the appropriate function based on the pixelarea's tag returning a pointer to that function. They always take into consideration the precision, sometimes also format and has_alpha. It's easy to see which row functions they refer to pointing to apply_mask_to_alpha_channel_row_bfp pointing to replace_row_bfp sets a whole bunch of row functions that corresponds to the layer modes, such as screen, difference... the following type-dependent functions are implemented directly in paint_funcs_area.c instead of paint_funcs_row_* (Why??) wraps up all of rle_row_bfp,gaussian_blur_row_bfp and gaussian_curve_bfp just added the new bfp image types apart from the functions, added a new array render_funcs_bfp just added new menu option for "image/bfp" to image_entries array new select case to set point to seed_fill_bfp add sensitivity setting for BFP conversion entry in image/bfp adds a given value to all pixels in a row of width w and starting at x,y I am not entirely certain this is implemented correctly, since I didn't study the brightness funcs in detail. It seems to have the right effect, thought pre-calculates function vales into a LUT added select case. the select seems to be unnecessary, though: select(precision) {case PRECISION_BFP: precision=5}, but PRECISION_BFP=5, anyway, so we could just assign add radio button to the interface !!MISSING!! only exist for u8 and u16, didn't know whether to implement for bfp the code is exactly the same for all types!? pre-calculates function vales into a LUT why do all types apart from float use 2*gamma and 2*expose in gamma/exposure funcs, but float doesn't? pre-calculates function vales into a LUT new select case to set the string formatting of BFP value apart from functions add new array ui_strings_bfp[4][8] which contains strings for min and max values pre-calculates function vales into a LUT added new plug-in image types for bfp introducing PRECISION_BFP tag-value new image types new select case for number of bins new select case to set number of bins