06-10-2021, 06:47 AM
Couple of quick points re: choosing photos to study from.
That portrait photo you chose is lit very ambiently, with little to no observable cast shadows, and even form shadows are diminished.
It is easier to dial up the difficulty progressively and use photos with more direct lighting and observable cast and form shadows first. Straight on or slightly angled poses are generally harder than 3/4 or profile. Try and go for the easier option next time. The orc one was a better lighting choice. Another question, do you reduce the contrast in the ref during your studies or is that just for presenting?
Young pretty women are also the hardest to get a good likeness of because the features are so subtle. Because of that, we tend to overexagerate midtone values more, one example being the bridge of her nose, which doesn't need to be delineated all the way down the form. Squint at your reference to reduce value information.
You're not doing too bad, but again change up the age/gender/lighting more. I also dont think doing 20 unfinished drawings of one poor/difficult photo ref is efficient, and smacks of obsession, not targeted learning.
And I will have to make yet another push for doing actual still life or studies from life. It's amazing for me to see how few people here actually bother to do actual studies from life, even just basic primitive shapes under lighting can be so incredibly useful for almost every fundamental. Oh well.
Draw the whole head to judge scale easier, floating features are not very easy to judge. You're not really making much changes between each one i.e making the same mistakes each time.
I'd say do the study, figure out where it's off. Do some other studies related to those off things but using different references (eg eye placement/shape, overall symmetry, edge control, value grouping, or whatever you identify needed work) then after a while go back and do another one of the initial study. That would be better than brute forcing numerous identical studies on one image repeatedly for days. Probably will be more engaging too I imagine.
And regarding using grids. I wouldn't. It's perhaps fine to quickly check where you went off after the fact, but don't rely on them for your layin drawings or through the whole process. It's a crutch that you don't want to get used to.
Edit: Hair is best dealt with volumetrically first, meaning putting in largest value masses first, then progressively refining details and strands where / if needed. Which also means you have to understand the importance of the notion of value grouping/massing. Most of the orc's hair can be grouped into one dark value/shadow area except for where the light hits it from top right. The fact you could not see this major element, makes me think you aren't considering value grouping of the whole or breaking down the values you see in a more general sense first. Again I think you're going too detail focused too quickly or you gradually creep up on the values, feature by feature rather than overall. So as a general philosophy, perhaps read up a bit on value grouping.
That portrait photo you chose is lit very ambiently, with little to no observable cast shadows, and even form shadows are diminished.
It is easier to dial up the difficulty progressively and use photos with more direct lighting and observable cast and form shadows first. Straight on or slightly angled poses are generally harder than 3/4 or profile. Try and go for the easier option next time. The orc one was a better lighting choice. Another question, do you reduce the contrast in the ref during your studies or is that just for presenting?
Young pretty women are also the hardest to get a good likeness of because the features are so subtle. Because of that, we tend to overexagerate midtone values more, one example being the bridge of her nose, which doesn't need to be delineated all the way down the form. Squint at your reference to reduce value information.
You're not doing too bad, but again change up the age/gender/lighting more. I also dont think doing 20 unfinished drawings of one poor/difficult photo ref is efficient, and smacks of obsession, not targeted learning.
And I will have to make yet another push for doing actual still life or studies from life. It's amazing for me to see how few people here actually bother to do actual studies from life, even just basic primitive shapes under lighting can be so incredibly useful for almost every fundamental. Oh well.
Draw the whole head to judge scale easier, floating features are not very easy to judge. You're not really making much changes between each one i.e making the same mistakes each time.
I'd say do the study, figure out where it's off. Do some other studies related to those off things but using different references (eg eye placement/shape, overall symmetry, edge control, value grouping, or whatever you identify needed work) then after a while go back and do another one of the initial study. That would be better than brute forcing numerous identical studies on one image repeatedly for days. Probably will be more engaging too I imagine.
And regarding using grids. I wouldn't. It's perhaps fine to quickly check where you went off after the fact, but don't rely on them for your layin drawings or through the whole process. It's a crutch that you don't want to get used to.
Edit: Hair is best dealt with volumetrically first, meaning putting in largest value masses first, then progressively refining details and strands where / if needed. Which also means you have to understand the importance of the notion of value grouping/massing. Most of the orc's hair can be grouped into one dark value/shadow area except for where the light hits it from top right. The fact you could not see this major element, makes me think you aren't considering value grouping of the whole or breaking down the values you see in a more general sense first. Again I think you're going too detail focused too quickly or you gradually creep up on the values, feature by feature rather than overall. So as a general philosophy, perhaps read up a bit on value grouping.