| | <<<Back 1 day (to 2011/12/28) | 2011/12/29 |
| henrys | ray_laptop:I doubt marcos is going to be around today I don't think we need a meeting. | 17:22.25 |
| ray_laptop | henrys: sorry -- I wasn't paying attention. I agree that a meeting is rather pointless. | 17:27.58 |
| | nobody (except maybe cust 532) expects much to get done this week. :-) | 17:28.29 |
| | (and my wife) | 17:28.56 |
| Robin_Watts | I hate transparency blending. | 17:29.24 |
| henrys | we should give the wives customer numbers. | 17:29.25 |
| Robin_Watts | (That's nothing to do with the conversation, just I had to say it) | 17:29.43 |
| ray_laptop | Robin_Watts: I don't blame you a bit. IMHO, Adobe went a bit "over the top" in the PDF transparency functionality | 17:30.38 |
| Robin_Watts | ray_laptop: Ghostscript seems to need WAY too much information. | 17:31.18 |
| | or at least, it needs way more than my tiny mind can comprehend. | 17:31.33 |
| henrys | so all this stuff works in mupdf and is simpler? | 17:32.24 |
| ray_laptop | Robin_Watts: the business of group alpha, and shape alpha confuses me. | 17:32.41 |
| Robin_Watts | henrys: No, that's the flaw of course. | 17:32.42 |
| ray_laptop | Robin_Watts: by 'too much info' are you referring to the number of planes ? | 17:33.18 |
| Robin_Watts | Yeah. I understand colour planes + alpha + shape. | 17:33.41 |
| | I don't follow group shape at all. | 17:33.57 |
| | group alpha, sorry. | 17:34.02 |
| ray_laptop | Robin_Watts: not sure, either, but I think it is required for knockout support | 17:34.21 |
| Robin_Watts | ray_laptop: That'd be a lot more convincing if gs got knockout right :) | 17:34.41 |
| ray_laptop | but don't follow me -- I'm lost | 17:34.48 |
| Robin_Watts | But this example runs through gs without using group alpha, as far as I can tell. | 17:35.17 |
| | And yet, I can't make mupdf match what gs does. | 17:35.26 |
| ray_laptop | Robin_Watts: you need mvrhel (or maybe raph) | 17:35.47 |
| | Robin_Watts: if the group alpha is never used, it may have just been inherited from 'libart' (Raph Levien's original foundation for the transparency in GS) | 17:37.22 |
| Robin_Watts | I won't swear it's not used. | 17:38.05 |
| ray_laptop | and just like mupdf, GS transparency effort was focused on the more common uses and never really got finished. In fact it languished for quite a while until mvrhel (and me a bit) dove into it | 17:39.26 |
| Robin_Watts | But the pdf14 stuff does really wierd stuff; like compositing stuff along the way, then 'uncompositing' before a blend. | 17:39.41 |
| | That just can't be sane, surely ? | 17:39.48 |
| | I moved mupdf to do (pretty much) the same as gs does last time I looked at this. | 17:40.14 |
| ray_laptop | Robin_Watts: I don't grok 'uncopositing' | 17:40.57 |
| Robin_Watts | ray_laptop: It takes a plane, and 'subtracts' the background from it (i.e. it removes the background from the plane) prior to blending with it. | 17:42.01 |
| | Surely it would be saner just not to have blended with the background in the first place? | 17:42.13 |
| ray_laptop | iirc, mupdf and gs handled alpha and color planes a bit differently -- something having to do with 'pre-multiplication' iirc | 17:42.30 |
| Robin_Watts | ray_laptop: Oh yes, but this is independent of whether we use premultiplied form or not. | 17:42.49 |
| ray_laptop | Robin_Watts: I agree, but don't understand why it may or may not be useful for some strange case (isolated, non-isolated, knockout, funky blend modes, etc.) | 17:46.16 |
| | I have to change venues. I'll be back on line in a bit. (not that I've been much help anyway) | 17:47.15 |
| Robin_Watts | My gut feeling is that for isolated cases, we never put the background in in the first place - then you never need to take it out. | 17:47.23 |
| ray_laptop | or even any help | 17:47.28 |
| Robin_Watts | And for non-isolated cases, you put the background in to start with. | 17:47.38 |
| | Then (I think) we never need to uncomposite. | 17:47.51 |
| ray_laptop | oops. phone call from cust 532 | 17:47.57 |
| Robin_Watts | tor8: ping! | 18:24.19 |
| | Well... I have l16.pdf giving the right results. | 18:26.27 |
| | So I'm going to call that success. | 18:26.42 |
| | Forward 1 day (to 2011/12/30)>>> | |
IRC Logs