发信人: Gatchaman (科学忍者队), 信区: ANSI
标 题: Re: 新作 & 探讨
发信站: 日月光华 (2010年04月18日12:22:51 星期天), 站内信件
同意这篇。
软件的确并非一无是处,特别是在真人色块的处理上很适合,
不过要转得好还是取决于原图本身,
否则通常仍会有一些纷乱的光影存在,细部也不明确。
另外则是,转图前对图片的处理能力也很重要,不光只是调高对比度就行的。
真人描摹个人认为还是在于本身需要有一些手绘素描基础,
才能有一定的把握与修正的能力,
想加强这方面的作品,其实不如从实际素描开始,
再不然至少也要多观察一些人体素描方面的图。
【 在 Castellio (percyshelley|Right to Heresy) 的大作中提到: 】
: 下面一幅是我刚画的东西。真人色块让我感到前所未有的棘手,主要原因是其复杂的光
: 影过渡与变化。与动漫不同,真人的颜色过渡非常不明显,让人无法判断阴影与本色的
: 分界线应该放在哪里。
: 我曾经试图将原图调高对比度,根据脸部颧骨、鼻梁等关键部位来确定明暗,却发现无
: 法同原图对应起来,变得非常古怪。
: 最后还是参考了转图的一些阴影走向。发现我现在的水平,完全无法企及转图的精确度
: 。我觉得动漫作品的绘制相对自由,即使有比较大的形变,基本特征抓住了,就八九不
: 离十。但绘制真人,真是差之毫厘,失之千里,甚至觉得还是跑程序比较能抓住神韵。
: 个人认为,跑程序并非一无是处。有时候程序转出来的图,一些细节处理能带来很大的
: .................(以下省略)
--
─ ▆ ﹨ ───────────── ﹨▂▄▅▅▅▄▃▁◤▄ ◣▃ ◢
███ 科学忍者队 ◤◤ ◤ ◣ ◣ ◥ ▆▃◤▊
█ ▉ ▃▅▇ ▅▇█ ▂ ∕ ◢ ▄▂ ▁ ◥◣ \ ◣ ◥◤ ▍ ▍
█▊▋│▎ ▅█▅ ▂ ▆ █ ▄ ◢ ∕◢◤▂ ◤ ◥▇ ◣ ▌
█▊█ ▋ ▎▏▅█▂ ◥◣ ▆◤▊| ▊▊◢˙ ◣\ 〃◢˙ ◣◣▊ ▎ ▕▌
──────────◥ ───── │▊ ▆▅▆/ ▆▅▆ \▎ ▍ ▉
※ 来源:·日月光华 bbs.fudan.edu.cn·[FROM: 210.192.236.*]
Sender: Gatchaman Science Ninja Team, message area: ANSI
Title: Re: New Work & Discussion
Sending site: Sun and Moon Guanghua April 18, 2010 12:22:51 Sunday, site mail
Agree with this article
The software is indeed not without its merits. It is especially suitable for processing real-life color blocks.
However, whether the conversion is good still depends on the original image itself.
Otherwise, there will usually still be some chaotic light and shadow, and the details will not be clear.
In addition, the ability to process the image before converting it is also very important. It is not just about raising the contrast.
Personally, I think that tracing a real person requires some basic hand-drawing sketching skills.
Only in order to have a certain degree of grasp and ability to correct
If you want to strengthen your work in this area, it’s better to start with actual sketching.
Otherwise, you should at least look at some more drawings of human figures.
In Castellio percyshelley Right to Heresy's masterpiece:
: The following picture is something I just painted. The real color block is more difficult for me than ever. The main reason is its complicated light.
: Shadow transition and change. Different from animation, the color transition of real people is very unobvious, making it impossible to judge the shadow and true color.
: Where should the dividing line be placed?
: I once tried to increase the contrast of the original image to determine the light and dark based on key parts of the face such as the cheekbones and bridge of the nose, but found that it was useless.
: When the method corresponds to the original picture, it becomes very weird.
: Finally, I referred to some shadow directions of the transferred image and found that my current level is completely unable to match the accuracy of the transferred image.
: I think the drawing of animation works is relatively free. Even if there are relatively large deformations, the basic characteristics are captured and it’s almost impossible.
: Li Shi But when it comes to drawing a real person, it’s really just a hair’s breadth and a thousand miles away. I even think it’s better to run a program to capture the charm.
: Personally, I think that running a program is not useless. Sometimes some details of the pictures transferred from the program can bring great benefits.
: ............. omitted below
﹨ ﹨
Science Ninja Team
Source: Sun Moon Guanghua bbs.fudan.edu.cn [FROM: 210.192.236.*]