User talk:Yahastu

Speedy deletion of He Jinbao
Hello Yahastu, this is a message from an automated bot to inform you that the page you created on May 6 2007, He Jinbao, has been marked for speedy deletion by User:Papaursa (page has mainspace links, and 42 edits). This has been done because the page seems to be about a person, band, club, company, or web content, but it does not indicate how or why the subject is important or significant (see CSD). If you think the tag was placed in error, please add " " to the page text, and edit the to explain why the page should not be deleted. If you have a question about this bot, please ask it at User talk:SDPatrolBot II. If you have a question for the user who tagged the article, see User talk:Papaursa. Thanks, - SDPatrolBot II (talk) on behalf of 05:46, 18 May 2010 (UTC)

File:Fire-breather CIELAB L*.jpg
On the HSL_and_HSV page you have extracted various lightness components for the Fire-breather image. It appears to me that the image you have shown for CIELAB L* is incorrect. After some investigation I think that the image was computed by using Rec. 708 equation 0.2126*R + 0.7152*G + 0.0722*B, except that it was used with the gamma-compressed (R'G'B') components instead of the linear RGB components. The actual image should be much darker and look significantly different from the Luma version. Yahastu (talk) 03:52, 4 May 2011 (UTC)


 * It’s maybe slightly tricky to call it L*, though I think the label is reasonable. The image is necessarily an sRGB image, because JPEG doesn’t support CIELAB space. So what it shows is a grayscale sRGB image in which each pixel has the same L* as the corresponding pixel in the original image. What you thought I was trying to create was instead a file wherein the pixels literally took on the values of L* in the original image; this of course would appear different (than both the image linked above and the original color image), because L* and the sRGB gamma function are somewhat different.


 * Because L* is a bijective function of luminance Y, it would be equally valid to say that each pixel in the resulting sRGB image has the same Y values as the corresponding pixel in the original. There might be some way of describing exactly what is shown that wouldn’t lead to potential confusion, but to be honest I can’t think of one: indeed all the alternative captions I can think of are likely to increase confusion. What I’m trying to show is that using L* as a lightness dimension better preserves perceived lightness relationships between colors than using one of the other dimensions described.


 * If you compare the “Luma” and “L*” images, you’ll notice that they are substantially different in highly saturated regions like the man’s face and clothes.


 * Hope that helps, jacobolus (t) 09:18, 4 May 2011 (UTC)


 * So to clarify, you first extracted the raw gamma-compressed R'G'B' from the color JPEG, then decoded the gamma to get linear RGB, then computed luminance as per. Rec. 708 above, and then re-encoded the gamma before saving as a grayscale JPEG, correct? Yahastu (talk) 13:58, 4 May 2011 (UTC)


 * I think actually what I did is took a file into Photoshop, converted to CIELAB, stipped out the a* and b* channels, and converted back to sRGB. But that basically amounts to the same thing as taking the sRGB file, decoding the gamma, adding the weighted components together, and then re-gamma-encoding. –jacobolus (t) 20:45, 4 May 2011 (UTC)