Re: [swfmill] prerelease 0.2.11.21
FYI- I created the test swfs with flash 8. I'm not sure what a convolution filter is, but the crash case was created in flash 8, so if there is such a filter in there, then it was also created with flash 8. Also, I was trying to narrow down the filter crash case, and I noticed that, in addition to colors in filters (there can be any number in gradient filters), there is also an "adjust color" filter that allows setting of "hue, saturation, contrast, brightness". This did seem to be causing problems, but not by itself, and not consistently, but I am guessing it is always parsed incorrectly, and only sometimes causes the error. Maybe this is the 18 byte mystery data blob? Would you like an example swf with just the "adjust color" filter applied? Regards, Hudson ___ swfmill mailing list swfmill@osflash.org http://osflash.org/mailman/listinfo/swfmill_osflash.org
Re: [swfmill] prerelease 0.2.11.21
Also... which is caused by the ColorTransform filter in your toimport_filter.swf. I > can parse the swf fine if i reduce the expected size of the color matrix > from 20 to 18. But ColorMatrices should have 20 values, no? According to the official specs, the ColorMatrix filter consists of (in addition to the standard tag header) an unsigned byte specifying the filter type (0x06 in the case of ColorMatrix) followed by 20 FLOAT records (which is a 32-bit single-precision value). If it helps I'm getting a std::bad_alloc exception thrown when trying to swf2xml the toimport_filters.swf file. Steve -- Steve Webster http://dynamicflash.com ___ swfmill mailing list swfmill@osflash.org http://osflash.org/mailman/listinfo/swfmill_osflash.org
Re: [swfmill] prerelease 0.2.11.21
Hey Dan, I think i found the problem with toimport.swf, seems we'll have to parse another byte if cacheAsBitmap is used (does it have different modes?). But before i prerelease this, i'd like to find out more about the other problem, Interesting. This extra byte is certainly not in the Adobe-provided docs. Still, it seems to work, and I don't know why I didn't spot this during development. which is caused by the ColorTransform filter in your toimport_filter.swf. I can parse the swf fine if i reduce the expected size of the color matrix from 20 to 18. But ColorMatrices should have 20 values, no? According to the official specs, the ColorMatrix filter consists of (in addition to the standard tag header) an unsigned byte specifying the filter type (0x06 in the case of ColorMatrix) followed by 20 FLOAT records (which is a 32-bit single-precision value). How did you generate that swf? with flash? how many values can you set there? also, it might be good to do tests for the other filters. Convolution uses a matrix of dynamic size... Steve, have you tested these at the time? Yes I did, with a test SWF similar to Hudson's. However, if I remember correctly you can't apply a convolution filter using the Flash 8 IDE so I was unable to properly test this. Presumably this filter exists in the SWF specs for 3rd party SWF-generating tools like swfmill, since the only way to apply it to a Flash SWF is to use ActionScript. and even further on, swfmill and the osflash community at large could profit from a set of well-crafted test cases that cover most of what the flash player (or players, including gnash et al and earlier versions) do-- if possible in a clearly visible way so pass/fail could be detected automatically by comparing screenshots... I agree that this would be very useful. Not something I have time for, but definitely useful. When implementing the filter support I spent a lot of time manually examining screenshots. Steve -- Steve Webster http://dynamicflash.com ___ swfmill mailing list swfmill@osflash.org http://osflash.org/mailman/listinfo/swfmill_osflash.org
Re: [swfmill] prerelease 0.2.11.21
I created the swfs with Flash (I get swfs from a designer who uses Flash). The blendMode type is a drop-down selection, and the cacheAsBitmap is a checkbox, independent of the blendMode setting (i.e., you can have cacheAsBitmap set with no blendMode). Maybe it uses a byte for that Boolean setting on it's own? For the filter issue, I added a couple of filters with no error, so I added one of each type and then I got the error. Maybe color matrices are 18 bytes within a filter... would there be something in a color matrix that would already be specified by the filter? I have to admit, I really have very little knowledge of the byte code breakdown of swfs :-) Flash let's you add a number of filters to one movieClip, I added about 7, but there is no indication of a limit. Would it help to make a SWF with a clip for each different type of filter? BTW- I had no good reason to have a movieClip with 7 filters on it (although you never know what a designer might come up with ;-) , I was just testing swfmill at that point... And I'm not sure if it was just one of the filters that caused the problem or that there were so many... Regards, Hudson On 1/2/07, daniel fischer <[EMAIL PROTECTED]> wrote: > "Hudson Ansley" <[EMAIL PROTECTED]> (on Tue, 2 Jan 2007 11:35:12 -0500): > > > I've posted two example SWFs with issues in a zip file at: > > http://hlamedia.com/swfmillPlaceObjectError2.zip > > Thanks Hudson, the files are perfectly small and crashing. > > I think i found the problem with toimport.swf, seems we'll have to parse > another byte if cacheAsBitmap is used (does it have different modes?). But > before i prerelease this, i'd like to find out more about the other problem, > > which is caused by the ColorTransform filter in your toimport_filter.swf. I > can parse the swf fine if i reduce the expected size of the color matrix from > 20 to 18. But ColorMatrices should have 20 values, no? > > How did you generate that swf? with flash? how many values can you set there? > > also, it might be good to do tests for the other filters. Convolution uses a > matrix of dynamic size... Steve, have you tested these at the time? > > and even further on, swfmill and the osflash community at large could profit > from a set of well-crafted test cases that cover most of what the flash > player (or players, including gnash et al and earlier versions) do-- if > possible in a clearly visible way so pass/fail could be detected > automatically by comparing screenshots... > > -dan > > > -- > http://0xDF.com/ > http://iterative.org/ > > ___ > swfmill mailing list > swfmill@osflash.org > http://osflash.org/mailman/listinfo/swfmill_osflash.org > ___ swfmill mailing list swfmill@osflash.org http://osflash.org/mailman/listinfo/swfmill_osflash.org
Re: [swfmill] prerelease 0.2.11.21
"Hudson Ansley" <[EMAIL PROTECTED]> (on Tue, 2 Jan 2007 11:35:12 -0500): > I've posted two example SWFs with issues in a zip file at: > http://hlamedia.com/swfmillPlaceObjectError2.zip Thanks Hudson, the files are perfectly small and crashing. I think i found the problem with toimport.swf, seems we'll have to parse another byte if cacheAsBitmap is used (does it have different modes?). But before i prerelease this, i'd like to find out more about the other problem, which is caused by the ColorTransform filter in your toimport_filter.swf. I can parse the swf fine if i reduce the expected size of the color matrix from 20 to 18. But ColorMatrices should have 20 values, no? How did you generate that swf? with flash? how many values can you set there? also, it might be good to do tests for the other filters. Convolution uses a matrix of dynamic size... Steve, have you tested these at the time? and even further on, swfmill and the osflash community at large could profit from a set of well-crafted test cases that cover most of what the flash player (or players, including gnash et al and earlier versions) do-- if possible in a clearly visible way so pass/fail could be detected automatically by comparing screenshots... -dan -- http://0xDF.com/ http://iterative.org/ ___ swfmill mailing list swfmill@osflash.org http://osflash.org/mailman/listinfo/swfmill_osflash.org
Re: [swfmill] prerelease 0.2.11.21
I've posted two example SWFs with issues in a zip file at:
http://hlamedia.com/swfmillPlaceObjectError2.zip
One generates the placeObject3 error ("toimport.swf"). This has a
library item with a blendmode *and* "cache bitmap" checked (the same
settings without cache bitmap does not generate an error).
Microsoft Windows XP [Version 5.1.2600]
C:\jobs\tests\swfmill placeObject error2>swfmill -v simple
application.xml index.swf
swfmill 0.2.11.21
Reading from application.xml
WARNING: end of tag PlaceObject3 is @162, should be @163
Writing SWF to index.swf
SWF saved to index.swf (70 bytes).
The second causes swfmill to end unexpectedly. This has a library item
with several bitmap filters applied.
>
C:\jobs\tests\swfmill placeObject error2>swfmill -v simple
application_filter.xml index.swf
swfmill 0.2.11.21
Reading from application_filter.xml
This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information.
<<<
I hope the example files are helpful.
Regards,
Hudson
___
swfmill mailing list
swfmill@osflash.org
http://osflash.org/mailman/listinfo/swfmill_osflash.org
Re: [swfmill] prerelease 0.2.11.21
Nice, that takes care of a bunch of placeObject3 errors I was getting. However, I am still getting some, so I think I will make a test SWF with a bunch of new Flash 8 features to see what that will turn up. Also, is there a way to have the swfmill output list the line in the XML input when given a warning (or error)? That would help me determine which imported SWF is generating the error. Currently I am just seeing a warning like this: WARNING: end of tag PlaceObject3 is @99553, should be @99554 (this is swfmil running through an ant process in eclipse) Regards, Hudson PS- a haxe port would be cool. I've been meaning to make that leap myself :-) On 1/2/07, daniel fischer <[EMAIL PROTECTED]> wrote: > > hey list, > > sorry for the delay; Steve's fix for #16 is in prerelease 0.2.11.21, > see > http://swfmill.org/trac/browser/trunk/NEWS for a changelog. > ___ swfmill mailing list swfmill@osflash.org http://osflash.org/mailman/listinfo/swfmill_osflash.org
[swfmill] prerelease 0.2.11.21
hey list, sorry for the delay; Steve's fix for #16 is in prerelease 0.2.11.21, see http://swfmill.org/trac/browser/trunk/NEWS for a changelog. 2007 will see a release for sure and probly a haxe port :) all the best to yous all. -dan -- http://0xDF.com/ http://iterative.org/ ___ swfmill mailing list swfmill@osflash.org http://osflash.org/mailman/listinfo/swfmill_osflash.org
