SVG Working Group Teleconference -- 04 Apr 2013

<trackbot> Date: 04 April 2013

calling in in a moment

<Tav> yes

<scribe> scribeNick: krit

tecon time

heycam: sent a mail with possible options

Tav: what about switching bi-weekly

heycam: thought about that

krit: might be confusing

heycam: what about 6AM?

Tav: would work for me

heycam: it is 9PM PST

richardschwerdtfeger: well, I am in the middle
... would be 12 in the morning

heycam: should we have 60min instead?

krit: yes

heycam: we might tighten up conversation

shepazu: if we have a frequent time to call in, it is managable

krit: agree

heycam: I put up a doodle and we figure it out

image-rendering property

Tav: we experimented with inscaling and outscaling
... I looke in CSS and saw that it has a crisp edges proposal
... wich seemed to be a good idea
... so if you scale, the pixels are preserved
... not filtered and get blurry

<heycam> http://dev.w3.org/csswg/css-images/#image-rendering

Tav: in my email there is a lingk to the draft

tav reads the proposal

shepazu: my suggestion SVG should simply do as if the image would be in HTML

krit: image-rendering was implement in SVG already
... with the same purpose

Tav: well, you could choose for quality or speed

krit: I think this should move to CSS WG
... it also does what Tav wants

Tav: we should say what it means for SVG

krit: not really

heycam: ok so we should reference the CSS spec.

krit: I think CSS should override the SVG defintion and we do not reference CSS IMage as long as it is not in a further step

<ed> typically image-rendering='optimizeSpeed' in svg content means "please use nearest-neighbor", and 'optimizeQuality' means use something better than nearest-neighbor (typically bilinear)

<ed> not sure if content depends on it, but 'auto' is different

shepazu: we cold just drop it from SVG2 and let CSS keep it up
... if someone wants it, he can use SVG1.1 defintion

Tav: there should be a notification

krit: can we put this note in the appendix?

Tav: it would be one step more for people to get the necessary information

shepazu: what about a compromise where we have a detailed appendix and short notes in the spec referencing the Appendix

heycam: I thought about that as well about all referenced specs

Tav: I just want to have a short note

heycam: fine with me

ed: what about the degault value

krit: it is still auto

ed: sure but I want to be able to optimise speed
... auto means smoothing, which is undesirable in at least all 'optimizeSpeed' content I've made

krit: that is the default anyway

<ed> crisp-edges would probably be what I'd expect optimizeSpeed to be resolved as

<ed> or perhaps as 'pixelated'

krit: I think we should move discussion to www-style and argue there if there are concerns

shepazu: optimization may have very different effects even between implementations. I think this property does not influence existing content a lot
... I think we should let the CSS WG deal with it

heycam: it is worth bringing it up
... to inform authors about optimized speed

krit: I am fine with leaving it now

heycam: we leave it and add a note that CSS WG is working on it. If CSS WG is quicker, we remove it and add a reference to it

<scribe> ACTION: tav to add a note to image-rendering property that CSS WG is dealing with a new version of it [recorded in http://www.w3.org/2013/04/04-svg-minutes.html#action01]

<trackbot> Created ACTION-3480 - Add a note to image-rendering property that CSS WG is dealing with a new version of it [on Tavmjong Bah - due 2013-04-11].

<Tav> https://launchpadlibrarian.net/136181479/scaling_test.svg

Tav: ther is a nother issue
... open the image i posted in the mail and change the zoom

heycam: that is the bilienar filtering that you see

Tav: I think it is actually the color spacwe
... one that you see is sRGB, the other is linear

krit: isn't it an implementation issue

Tav: yes, but all three browsers do that because they use sRGB
... And I don't think that this is the right thing to do
... in the future when people use GPU it will be easy for themm to switch color spaces

<ed> the svg one is using a <pattern>, is it different if you use <image xlink:href="data:image/svg+xml..." />?

cabanier: it depends more on how you scale. I do not think that switching the color space is correct

shepazu: I don't think that is a SVG issue, but more CSS

cabanier: I do not think that you change color space on scaling
... transforms and color space are different thing

Tav: if I zoom in the color should not change

krit: I think this is more an issue of antialiasing, not color space

<shepazu> (is this different for SVG vs HTML? if not, why are we talking about it in an SVG telcon, vs discussing it on www-style?)

krit: there would definitely be a visible difference on changing color space on scaling and you can not know if you are scaling in a motion or not

cabanier: why specifying it in SVG if you think that is an issue in the implementation?

krit: we should move this to the mailing list and disucss it with the CSS WG

shepazu: I think we should not do our own color stuff, i think it should be in CSS

Tav: If SVG says it is something that we want to do, then we should drive it

shepazu: who is doing the color stuff in SVG

krit: Chris

cabanier: but Tab took over on CSS4 Color

<scribe> ACTION: Tav produce an exmaple, demonstrating how different color space interpolation effects image scaling [recorded in http://www.w3.org/2013/04/04-svg-minutes.html#action02]

<trackbot> Created ACTION-3481 - Produce an exmaple, demonstrating how different color space interpolation effects image scaling [on Tavmjong Bah - due 2013-04-11].

Tav: I saw the same issue on filters

krit: makes sense, but depends on the input

Tav: are browsers interestef in color interpolation?

cabanier: yes it is in there
... CSS4 Color

<heycam> http://dev.w3.org/csswg/css-color/#color-management

Tav: that would solve some problems that I see with meshes

cabanier: this is a property that switches between DeviceRGB and sRGB

krit: yes, that is what we have in WebKit
... in SVG we have color-interpoaltion that switches between sRGB and linearRGB

heycam: I think color-interpolation should go to CSS4 Color

krit: there is just one problem: SVG is sRGB by default. Implemeentd is DeviceRGB

cabanier: don't think CSS changes
... that

krit: I think it does, because it has auto, which is DeviceRGB

cabanier: oh

krit: in reality no one implements sRGB but DeviceRGB (all browsers and tools)

heycam: I don' think it says anything about using sRGB by default on images

krit: color-interpolation is not limited to images but effects everything

heycam: you are aying CSS should call it color-interpolation?

krit: yes, but we would need to add auto and would need to say that auto means sRGB
... or we say the truth and use DeviceRGB instead
... as every implementation does anyway

<scribe> ACTION: krit to ask CSS WG to use color-interpolation instead (as defined in SVG) with the addotion of auto [recorded in http://www.w3.org/2013/04/04-svg-minutes.html#action03]

<trackbot> Created ACTION-3482 - Ask CSS WG to use color-interpolation instead (as defined in SVG) with the addotion of auto [on Dirk Schulze - due 2013-04-11].

<scribe> ACTION: krit to figure out if we should use sRGB or DeviceRGB on SVG [recorded in http://www.w3.org/2013/04/04-svg-minutes.html#action04]

<trackbot> Created ACTION-3483 - Figure out if we should use sRGB or DeviceRGB on SVG [on Dirk Schulze - due 2013-04-11].

meeting with IDPF

s/IDPF/IDPF/

shepazu: I would not be able to go there
... IDPF is happeing at Keio Mita campus
... 25min appart by train
... agenda is about internationalization issues
... are any of you interested in this topic?

heycam: you said that they may have interest in SVG and want our opinion about that?

<richardschwerdtfeger> :-)

shepazu: I think the focus is on electrical publishing

richardschwerdtfeger: It is really important in MathML
... I'd like to have a open format in documetns
... I would like to see them SVG2
... which is important in the future
... ARIA and other accesibility things

shepazu: I am not convinced if the timing is right
... and especially in the context of this meeting
... the focus on internationalizing

resolve: we meet from monday to wednesday

shepazu: I will talk with the organizers
... another option is to invite them to one of our sessions

<birtles> by the way, I started adding some details for the Tokyo F2F to http://www.w3.org/Graphics/SVG/WG/wiki/F2F/Tokyo_2013, more hotel suggestions etc. to come

<heycam> cool

shepazu: why not invite some to a telcon

<scribe> ACTION: shepazu to invite IDPF people to a SVG telcon [recorded in http://www.w3.org/2013/04/04-svg-minutes.html#action05]

<trackbot> Created ACTION-3484 - Invite IDPF people to a SVG telcon [on Doug Schepers - due 2013-04-11].

CSS4 Color and merging SVG Color

krit: tab sugeested to pick up SVG color things to CSS4 color
... ICC color profiles have been in SVG1.1 already. Assuming that there are viewers with support for ICC, it is fine to have it in SVG2 for now. But LAB is new. For browsers it makes sense if it is part of CSS4 Color

heycam: I am more affraid on relying on future specs in SVG2
... so new stuff can go to CSS4 Color
... not sure about old studd

stuff

<richardschwerdtfeger> dropping

heycam: they are in SVG2 now, because they were in SVG Color 1.2
... and we decided to merge them again with SVG2
... I want to hear ChrisL s opinion first
... I was on removing everything which is in CSS3 Color
... In the future it makes sense to have everything in CSS Color
... Are you concerned they don't get implemented in browsers and will be removed anyway?

krit: yes

heycam: we can around that problem in CR

<heycam> we could have a new conformance class for "color managed implementation"

krit: I think we should collect data about active SVG viewer implementations
... I am saying that colors in general should be managed by CSS Color

heycam: I agree, but want Chris's input

- DRAFT -

SVG Working Group Teleconference

04 Apr 2013

Attendees

Contents

tecon time

image-rendering property

meeting with IDPF

CSS4 Color and merging SVG Color

Summary of Action Items

Scribe.perl diagnostic output