{"topic":{"id":13924,"title":"2.4 光源相机该怎么设置参数","slug":"topic","userId":2329,"categoryId":56,"createdAt":"2024-11-28 04:59:59.794668","bumpedAt":"2024-11-28 12:57:37.006209","views":259,"postsCount":3,"highestPostNumber":3,"likeCount":0,"closed":false,"pinned":false,"pinnedGlobally":false,"pinnedAt":null,"pinnedUntil":null,"isBanner":false,"excerpt":"我是用这些参数来构造光源处的相机，我尝试过把用场景相机的近平面、远平面乘上一些系数再加上一些偏移，但是效果都不太好，顶点的深度都会映射到一个非常小的范围里面，只有这么设置参数才会看到顶点的深度不会那么集中。如果直接用相机的近平面、远平面，顶点经过 MVP 变换之后深度都在 0.9，请问光源相机的视锥体取的很大有影响吗？打印 shadow_map 却发现都是无穷大，但是渲染出来的结果：其他参数结果是这样：","deletedAt":null,"isAnonymous":false,"isPrivate":false,"isMembersOnly":false,"category":{"id":56,"name":"计算机图形学","slug":"cg","color":"92278F"},"author":{"id":2329,"username":"Arcw","name":"void face","avatarUrl":"/uploads/image/9c/c7/9cc78443c9741385f89b03dd23dd4a504f973932bedd05666ed18e83a959c8a4.avif"},"realAuthor":{"id":2329,"username":"Arcw","name":"void face","avatarUrl":"/uploads/image/9c/c7/9cc78443c9741385f89b03dd23dd4a504f973932bedd05666ed18e83a959c8a4.avif"},"tags":[]},"posts":[{"id":43725,"postNumber":1,"raw":"![image|690x55](/uploads/image/09/a3/09a37854cd9c5dba5cdf00d2614cf3eeb8116c3d418d1c2aeee6c5093d9270bf.avif)\n我是用这些参数来构造光源处的相机，我尝试过把用场景相机的近平面、远平面乘上一些系数再加上一些偏移，但是效果都不太好，顶点的深度都会映射到一个非常小的范围里面，只有这么设置参数才会看到顶点的深度不会那么集中。如果直接用相机的近平面、远平面，顶点经过 MVP 变换之后深度都在 0.9，请问光源相机的视锥体取的很大有影响吗？\n打印 shadow_map 却发现都是无穷大，\n![image|613x500](/uploads/image/0e/d4/0ed4b35be4fad0a7e3b1ccc77fc5ac7251f78c0143ab3107315c7e6854807544.avif)\n但是渲染出来的结果：\n![image|563x500](/uploads/image/c2/ac/c2ace7ed0dad736c3ee4018ffef55c4a69a0da32301b8f2d3def02cf904e92cb.avif)\n其他参数结果是这样：\n![3d4887def5ed1b3a307bccb002989c3|604x500](/uploads/image/ad/84/ad848f53ae216dc15cae7455966766940f729d43c267343f7ff439ea45ce2734.avif)","cooked":"

$\"image\"$
\n我是用这些参数来构造光源处的相机，我尝试过把用场景相机的近平面、远平面乘上一些系数再加上一些偏移，但是效果都不太好，顶点的深度都会映射到一个非常小的范围里面，只有这么设置参数才会看到顶点的深度不会那么集中。如果直接用相机的近平面、远平面，顶点经过 MVP 变换之后深度都在 0.9，请问光源相机的视锥体取的很大有影响吗？
\n打印 shadow_map 却发现都是无穷大，
\n $\"image\"$
\n但是渲染出来的结果：
\n $\"image\"$
\n其他参数结果是这样：
\n $\"3d4887def5ed1b3a307bccb002989c3\"$

\n","createdAt":"2024-11-28 05:00:00.044638","updatedAt":"2024-11-28 05:00:00.044638","replyToPostNumber":null,"replyCount":0,"likeCount":0,"wordCount":188,"liked":false,"bookmarked":false,"polls":[],"reactions":[],"wiki":false,"version":1,"publicVersion":1,"deletedAt":null,"isAnonymous":false,"user":{"id":2329,"username":"Arcw","name":"void face","avatarUrl":"/uploads/image/9c/c7/9cc78443c9741385f89b03dd23dd4a504f973932bedd05666ed18e83a959c8a4.avif","admin":false,"moderator":false,"signatureRaw":null,"signatureCooked":null}},{"id":43732,"postNumber":2,"raw":"\ndepth_buffer 初值是 inf，说明这个地方没有任何东西，所以出现 inf 很正常\n其实深度映射集中问题也不大，你可以把光源相机渲染的深度图打出来看看实际是什么样，我的光源相机 far_plane 只设置成了 200.0f，看起来都够用。\n![undefined|665x500, 75%](/uploads/image/1a/ec/1aecbeb7b49008fa5e68d7a693a5a5739fe5d3de650855e22e0c8f1108fcd69a.avif)\n后面的图看起来有点像 shadow acne 的问题，需要调整 shadow bias，也可以使用根据光源和法线角度（注意法线插值问题）动态调整 bias 算法之类的。此外提高深度图分辨率也可以。\n但最主要是你第一张图 cube 上都看不到 cow 的影子，我觉得你应该先把影子效果做出来再看失真问题。","cooked":"

depth_buffer 初值是 inf，说明这个地方没有任何东西，所以出现 inf 很正常
\n其实深度映射集中问题也不大，你可以把光源相机渲染的深度图打出来看看实际是什么样，我的光源相机 far_plane 只设置成了 200.0f，看起来都够用。
\n $\"undefined\"$
\n后面的图看起来有点像 shadow acne 的问题，需要调整 shadow bias，也可以使用根据光源和法线角度（注意法线插值问题）动态调整 bias 算法之类的。此外提高深度图分辨率也可以。
\n但最主要是你第一张图 cube 上都看不到 cow 的影子，我觉得你应该先把影子效果做出来再看失真问题。

\n","createdAt":"2024-11-28 06:54:46.067028","updatedAt":"2024-11-28 06:54:46.067028","replyToPostNumber":null,"replyCount":0,"likeCount":0,"wordCount":202,"liked":false,"bookmarked":false,"polls":[],"reactions":[],"wiki":false,"version":1,"publicVersion":1,"deletedAt":null,"isAnonymous":false,"user":{"id":3740,"username":"Lskyi","name":null,"avatarUrl":null,"admin":false,"moderator":false,"signatureRaw":null,"signatureCooked":null}},{"id":43740,"postNumber":3,"raw":"但是我打印出第一个 pass 光栅化三角形时候各个点的坐标（屏幕坐标系下的 x,y 和深度）\n![image|447x337](/uploads/image/d0/de/d0de6fe579ba02478f0739fb90df059a9f642fc09956e53c57063028eb9acf85.avif)\n\n发现好多个点的深度是一模一样的。\n\n而且调整过后发现 shadow_buffer 还是没有更新\n![image|690x279](/uploads/image/78/c1/78c1401213db0f40638f3083088a0273a1b8034c2a8768d82cfd2f4180911c55.avif)\n\n但是代码运行的时候有输出更新了 shadow_buffer\n![image|85x500](/uploads/image/1e/2c/1e2c45a0f1391bfdf491eeb11a4702abbd63c92126d1881f3eded48601b85224.avif)\n\n加上 bias 能消除条纹了\n![image|645x482](/uploads/image/70/21/7021af9996d19cff07c4266fb2422457117a5f2b666b1161178dcfcdbd45c16a.avif)\n\n请问是需要使用平行光来生成阴影才可以吗？","cooked":"

但是我打印出第一个 pass 光栅化三角形时候各个点的坐标（屏幕坐标系下的 x,y 和深度）
\n $\"image\"$

发现好多个点的深度是一模一样的。

而且调整过后发现 shadow_buffer 还是没有更新
\n $\"image\"$

但是代码运行的时候有输出更新了 shadow_buffer
\n $\"image\"$

加上 bias 能消除条纹了
\n $\"image\"$

请问是需要使用平行光来生成阴影才可以吗？

\n","createdAt":"2024-11-28 12:57:37.006209","updatedAt":"2024-11-28 12:57:37.006209","replyToPostNumber":null,"replyCount":0,"likeCount":0,"wordCount":117,"liked":false,"bookmarked":false,"polls":[],"reactions":[],"wiki":false,"version":1,"publicVersion":1,"deletedAt":null,"isAnonymous":false,"user":{"id":2329,"username":"Arcw","name":"void face","avatarUrl":"/uploads/image/9c/c7/9cc78443c9741385f89b03dd23dd4a504f973932bedd05666ed18e83a959c8a4.avif","admin":false,"moderator":false,"signatureRaw":null,"signatureCooked":null}}],"hasMore":false,"lastReadPostNumber":3,"hasUrlPostNumber":true}