[jHof]

Trying to harness value better and quicker. I seem todo alright with tradtional medium. I'm using these studies as a way to get around this learning curve with my wacom as well. The plastic on plastic is taking some getting use to.

I'm open for better ways of doing this. I just used photo referance of some Cow Boys. It's not my intent to trace or use the color selector but to eye the color selection instead.

I know there was that cube study way back when, just seemed so technical (and I'm sure very informative.) I just kinda wana dive in and maybe see how other people approach this subject.

(Didn't notice I had used both warm and cool greys on this one until after I zoomed back in, Whoop.)






Good idea? Wasting my time?

[sodiaz]

Great idea for a thread jHof =)

I'm no expert here but, aren't you supposed to use a colored reference ? And then try to eye ball the values from that. And when you are finished with the painting, you can desaturate the reference photo, to check if you got the values right. Anyways thats the way i do them =)

Here is one i did last night.

[jHof]

Sodiaz Nice work. Yea, that does seem like a good/better way todo it. I'll post my new ones up later tonight.

Looks like you could have went even darker for your darks. Kind of a middle to high key you have in there.

[Mikko K]

Sodiaz, if you're using photoshop, try changing your color mode from RGB to grayscale. That should give you the correct values from the color picture, desaturate doesn't discard the color information right away, but somehow separates the color channels from each other. So for example, same value but different hue/saturation might produce different results when desaturated.

[sodiaz]

Thanks jHof, jep you are right about the darks, I'm especially having a hard time getting green colors right.

Thanks Mikko, I didn't know that. That makes the result very different, some of my errors disappeared, hehe and a whole bunch of new errors showed up :p

[jfrancis]

Google rgb to luminance, or just visit this link:
http://www.scantips.com/lumin.html

...you'll see that the correct way to convert rgb to luminance is is compute the weighted average of r, g, and b as follows: 30% red + 59% green + 11% blue.

You CAN do that in photoshop, but not by sliding the desat slider down.

Photshop does something weird -- it treats fully saturated red, green, and blue as if they all desrve the same gray value.

If someone wants to make a Photoshop "action" that computes the weights correctly, it would be handy here.

[sodiaz]

Thanks for the tip, and link jfransis =) I'm gonna try that. This stuff is more complicated then i thought

[kaboom_racoon]

Ok, this has to be the dullest first post anyone has ever made, but atleast I brought a gift.

I�m really to crappy to post any pics yet, but since I know my way around PS I made this little action that mixes the channels and makes a greyscale pic out of your RGBphoto.

I used jfrancis values, but if you�d like to use different ones than just go to
Image>adjustments>channel mixer... and fool around with the sliders.

http://elva.host.sk/RGB-gray.zip

Ok, now I�ll slip back to lurkermode again.

[jfrancis]

That equation is the one that televisions use. You can see from this google search:

http://www.google.com/search?hl=en&ie=UTF-8&oe=UTF-8&q=rgb+to+luminance&btnG=Google+Search

...that it is the most commonly used equation for converting rgb to luminance

[sodiaz]

Thanks allot for taking the time to make the PS action, Kaboom and thanks again for the info jfrancis. At first i thought that it didn't work, because the level/value of the original and the converted pic. differed when i tested with the pipette. But it turns out, that it was because of the JPG compression just tested on some RAW images, and it works

[jHof]

Hey all, I didn't mean to abandon this thread yet. This is still an exersize I want to continue. Been busy college lately. SOON : )

[jfrancis]

You know what might be a good exercise for this thread? Adding the use of posterization (region banding) into the process -- like a color-by-number painting, but in grayscale.

Take a photograph and try to predict in advance into which of, say, 5 gray levels a given region will fall. Make a speedpaint of your predictions.

Turn the photo image into grayscale using your method of choice (as discussed above) and posterize it into 5 gray levels -- black, dark, mid, bright, and white

Maybe blur the grayscal photo a bit before posterizing it just to simplify the regions somewhat.

Check your predicted speedpaint against the actual one.

When you get good at 5 levels of posterization, try 7...

I think this could develop powers of comparison in a way that starts simple and builds with practice.