Talk:TSL color space
 | This article is rated Stub-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects: | ||||||||||
| |||||||||||
.png){kind=small}
Are TSL and HLS equivalent?
[edit]Are TSL and HSL/HLS/HSI the same colour-space? Is there any difference between 'Tint' and 'Hue'? The French equivalent of the HSL color space article is called Teinte saturation lumière (which roughly translates as Tint, Saturation, Luminance). If so, this should be merged into the HSL color space article. Ae-a 11:17, 13 July 2006 (UTC)
- by their mathematical definitions, they're different. But I'm not clear on what this space is or what it is used for. --jacobolus (t) 18:48, 19 September 2007 (UTC)
- In response to what it's used for: the only place I've seen "tint" before is on just about every tube-style television I've ever had. 70.234.248.201 (talk) 18:22, 3 June 2009 (UTC)Michael
- It's a color space facial recognition. The values aren't supposed to mean anything comprehensible, they're just supposed to make the distance in the color space a good approximation of human vs not. As far as I can tell, at least. The "tint" dimension is actually this weird periodic technicolor vomit thing. But all the humanish colors are roughly in the T=2/5 slice of the color space so good job, weird color space. 24.22.235.130 (talk) 04:03, 1 May 2014 (UTC)
- it's basically a polar version of gamma-corrected rg chromaticity centered at 1/3, 1/3, with TV Luma tacked on. the definition can probably be simplified with atan2. Artoria2e5 🌉 02:20, 24 June 2023 (UTC)
- It's a color space facial recognition. The values aren't supposed to mean anything comprehensible, they're just supposed to make the distance in the color space a good approximation of human vs not. As far as I can tell, at least. The "tint" dimension is actually this weird periodic technicolor vomit thing. But all the humanish colors are roughly in the T=2/5 slice of the color space so good job, weird color space. 24.22.235.130 (talk) 04:03, 1 May 2014 (UTC)
Where did the conversion from TSL to RGB come from?
[edit]I don't see this equation anywhere in the literature, and the equations given can't possibly be right. They don't include L anywhere as a term. I don't know the right equations, but it's not this. 24.22.235.130 (talk) 06:21, 29 April 2014 (UTC)
- Fixed. I was going to provide a visualization of the color space while I was at it, but you don't want it. It's a very ugly color space. 24.22.235.130 (talk) 03:47, 1 May 2014 (UTC)
- Where did you find the new equations? The cited article has different definitions, and the inverse conversion doesn't seem to return anything near the RGB values I put in. HairyFotr (talk) 18:02, 5 May 2014 (UTC)
- I just guessed until I got the same L value back out by converting twice. One of the other terms could still be wrong. And that's all implemented based on this page's version of the original equations; if they're wrong then so is the inverse. I can't find the inverse equation anywhere in the literature; I have no clue where the one I tried to correct came from. I'm also not sure if the topic is notable in the first place, since I only see TSL referenced in a handful of papers. 24.22.235.130 (talk) 15:28, 27 May 2014 (UTC)
- The formulas are still not correct. I put them in a spreadsheet and I'm getting weird values for RGB (as in between -2300 and 4000 for 255-based components). I put them in public google docs:
- Read/Write link (might be subject to vandalism): https://docs.google.com/spreadsheets/d/1pNVIEsxwG4j8FdHSIXbmobAhDlD_HbWYA7Yzwuh7Ah0/edit?usp=sharing
- Read-only link for reference: https://docs.google.com/spreadsheets/d/1aeX8yPvRmiM5JhScvGx2Ub9CV1Sh8CVC4Kho7flwsKo/edit#gid=162141974
- Jpmaterial (talk) 08:19, 17 January 2017 (UTC)
- Someone else found the same problem and got this other solution: https://github.com/colorjs/color-space/commit/8ecc284f0c244a82dda42acd0c12b5c85e5f0ee3 Artoria2e5 🌉 16:00, 25 June 2023 (UTC)
- The formulas are still not correct. I put them in a spreadsheet and I'm getting weird values for RGB (as in between -2300 and 4000 for 255-based components). I put them in public google docs:
- Indeed, 'tan' cannot invert 'arctan2'. The formula given discards the sign and can only reproduce colors where 'r prime' and 'g prime' are in the first quadrant. I found a correct formula using 'sin()' and 'cos()'. Even if a formula with tangent is possible, it is not desirable as tangent has singularity at pi. — Preceding unsigned comment added by 2A02:3100:18E6:6C00:6302:45EC:2CBD:E424 (talk) 15:17, 21 November 2024 (UTC)