/**
* ObjectRenderer constructor
- * Note the ShaderProgram constructor which compiles the shaders for GPU rendering (if they exist)
+ * Note we cannot compile the shaders in the ShaderProgram constructor
+ * because the Screen class needs to initialise GL first and it has a
+ * ShaderProgram member
*/
ObjectRenderer::ObjectRenderer(const ObjectType & type,
const char * vert_glsl_file, const char * frag_glsl_file, const char * geom_glsl_file)
/**
* Render using GPU
*/
-void ObjectRenderer::RenderUsingGPU()
+void ObjectRenderer::RenderUsingGPU(unsigned first_obj_id, unsigned last_obj_id)
{
- if (!m_shader_program.Valid())
- Warn("Shader is invalid (objects are of type %d)", m_type);
+ // If we don't have anything to render, return.
+ if (first_obj_id == last_obj_id) return;
+ // If there are no objects of this type, return.
+ if (m_indexes.empty()) return;
+ unsigned first_index = 0;
+ while (m_indexes.size() > first_index && m_indexes[first_index] < first_obj_id) first_index ++;
+ unsigned last_index = first_index;
+ while (m_indexes.size() > last_index && m_indexes[last_index] < last_obj_id) last_index ++;
+
m_shader_program.Use();
m_ibo.Bind();
- glDrawElements(GL_LINES, m_indexes.size()*2, GL_UNSIGNED_INT, 0);
+ glDrawElements(GL_LINES, (last_index-first_index)*2, GL_UNSIGNED_INT, (GLvoid*)(2*first_index*sizeof(uint32_t)));
}
-/**
- * Helper structuretransforms coordinates to pixels
- */
-
-ObjectRenderer::CPURenderBounds::CPURenderBounds(const Rect & bounds, const View & view, const CPURenderTarget & target)
-{
- Rect view_bounds = view.TransformToViewCoords(bounds);
- x = view_bounds.x * target.w;
- y = view_bounds.y * target.h;
- w = view_bounds.w * target.w;
- h = view_bounds.h * target.h;
-}
/**
* Default implementation for rendering using CPU
*/
-void ObjectRenderer::RenderUsingCPU(const Objects & objects, const View & view, const CPURenderTarget & target)
+void ObjectRenderer::RenderUsingCPU(const Objects & objects, const View & view, const CPURenderTarget & target, unsigned first_obj_id, unsigned last_obj_id)
{
Error("Cannot render objects of type %d on CPU", m_type);
+ //TODO: Render a rect or something instead?
}
/**
m_indexes.reserve(max_objects); //TODO: Can probably make this smaller? Or leave it out? Do we care?
// Initialise and resize the ibo (for GPU rendering)
+ m_ibo.Invalidate();
m_ibo.SetUsage(GraphicsBuffer::BufferUsageStaticDraw);
m_ibo.SetType(GraphicsBuffer::BufferTypeIndex);
m_ibo.Resize(max_objects * 2 * sizeof(uint32_t));
/**
* Rectangle (filled)
*/
-void RectFilledRenderer::RenderUsingCPU(const Objects & objects, const View & view, const CPURenderTarget & target)
+void RectFilledRenderer::RenderUsingCPU(const Objects & objects, const View & view, const CPURenderTarget & target, unsigned first_obj_id, unsigned last_obj_id)
{
for (unsigned i = 0; i < m_indexes.size(); ++i)
{
- CPURenderBounds bounds(objects.bounds[m_indexes[i]], view, target);
- for (int x = max(0, bounds.x); x < min(bounds.x+bounds.w, target.w); ++x)
+ if (m_indexes[i] < first_obj_id) continue;
+ if (m_indexes[i] >= last_obj_id) continue;
+ PixelBounds bounds(CPURenderBounds(objects.bounds[m_indexes[i]], view, target));
+ for (int64_t x = max((int64_t)0, bounds.x); x <= min(bounds.x+bounds.w, target.w-1); ++x)
{
- for (int y = max(0, bounds.y); y < min(bounds.y+bounds.h, target.h); ++y)
+ for (int64_t y = max((int64_t)0, bounds.y); y <= min(bounds.y+bounds.h, target.h-1); ++y)
{
int index = (x+target.w*y)*4;
target.pixels[index+0] = 0;
/**
* Rectangle (outine)
*/
-void RectOutlineRenderer::RenderUsingCPU(const Objects & objects, const View & view, const CPURenderTarget & target)
+void RectOutlineRenderer::RenderUsingCPU(const Objects & objects, const View & view, const CPURenderTarget & target, unsigned first_obj_id, unsigned last_obj_id)
{
+ //Debug("Render %u outlined rectangles on CPU", m_indexes.size());
for (unsigned i = 0; i < m_indexes.size(); ++i)
{
- CPURenderBounds bounds(objects.bounds[m_indexes[i]], view, target);
- for (int x = max(0, bounds.x); x < min(bounds.x+bounds.w, target.w); ++x)
- {
- int top = (x+target.w*max(0, bounds.y))*4;
- int bottom = (x+target.w*min(bounds.y+bounds.h, target.h))*4;
- for (int j = 0; j < 3; ++j)
- {
- target.pixels[top+j] = 0;
- target.pixels[bottom+j] = 0;
- }
- target.pixels[top+3] = 255;
- target.pixels[bottom+3] = 255;
- }
+ if (m_indexes[i] < first_obj_id) continue;
+ if (m_indexes[i] >= last_obj_id) continue;
+ PixelBounds bounds(CPURenderBounds(objects.bounds[m_indexes[i]], view, target));
+
+ // Using bresenham's lines now mainly because I want to see if they work
+ // top
+ ObjectRenderer::RenderLineOnCPU(bounds.x, bounds.y, bounds.x+bounds.w, bounds.y, target);
+ // bottom
+ ObjectRenderer::RenderLineOnCPU(bounds.x, bounds.y+bounds.h, bounds.x+bounds.w, bounds.y+bounds.h, target);
+ // left
+ ObjectRenderer::RenderLineOnCPU(bounds.x, bounds.y, bounds.x, bounds.y+bounds.h, target);
+ // right
+ ObjectRenderer::RenderLineOnCPU(bounds.x+bounds.w, bounds.y, bounds.x+bounds.w, bounds.y+bounds.h, target);
- for (int y = max(0, bounds.y); y < min(bounds.y+bounds.h, target.h); ++y)
- {
- int left = (max(0, bounds.x)+target.w*y)*4;
- int right = (min(bounds.x+bounds.w, target.w)+target.w*y)*4;
- for (int j = 0; j < 3; ++j)
- {
- target.pixels[left+j] = 0;
- target.pixels[right+j] = 0;
- }
- target.pixels[left+3] = 255;
- target.pixels[right+3] = 255;
-
- }
+ // Diagonal for testing (from bottom left to top right)
+ //ObjectRenderer::RenderLineOnCPU(bounds.x,bounds.y+bounds.h, bounds.x+bounds.w, bounds.y,target, C_BLUE);
+ //ObjectRenderer::RenderLineOnCPU(bounds.x+bounds.w, bounds.y+bounds.h, bounds.x, bounds.y, target,C_GREEN);
}
}
/**
* Circle (filled)
*/
-void CircleFilledRenderer::RenderUsingCPU(const Objects & objects, const View & view, const CPURenderTarget & target)
+void CircleFilledRenderer::RenderUsingCPU(const Objects & objects, const View & view, const CPURenderTarget & target, unsigned first_obj_id, unsigned last_obj_id)
{
for (unsigned i = 0; i < m_indexes.size(); ++i)
{
- CPURenderBounds bounds(objects.bounds[m_indexes[i]], view, target);
- int centre_x = bounds.x + bounds.w / 2;
- int centre_y = bounds.y + bounds.h / 2;
+ if (m_indexes[i] < first_obj_id) continue;
+ if (m_indexes[i] >= last_obj_id) continue;
+ PixelBounds bounds(CPURenderBounds(objects.bounds[m_indexes[i]], view, target));
+ int64_t centre_x = bounds.x + bounds.w / 2;
+ int64_t centre_y = bounds.y + bounds.h / 2;
- Debug("Centre is %d, %d", centre_x, centre_y);
- Debug("Bounds are %d,%d,%d,%d", bounds.x, bounds.y, bounds.w, bounds.h);
- Debug("Windos is %d,%d", target.w, target.h);
- for (int x = max(0, bounds.x); x < min(bounds.x+bounds.w, target.w); ++x)
+ //Debug("Centre is %d, %d", centre_x, centre_y);
+ //Debug("Bounds are %d,%d,%d,%d", bounds.x, bounds.y, bounds.w, bounds.h);
+ //Debug("Windos is %d,%d", target.w, target.h);
+ for (int64_t x = max((int64_t)0, bounds.x); x <= min(bounds.x+bounds.w, target.w-1); ++x)
{
- for (int y = max(0, bounds.y); y < min(bounds.y + bounds.h, target.h); ++y)
+ for (int64_t y = max((int64_t)0, bounds.y); y <= min(bounds.y + bounds.h, target.h-1); ++y)
{
- double dx = 2.0*(double)(x - centre_x)/(double)(bounds.w);
- double dy = 2.0*(double)(y - centre_y)/(double)(bounds.h);
- int index = (x+target.w*y)*4;
+ Real dx(2); dx *= Real(x - centre_x)/Real(bounds.w);
+ Real dy(2); dy *= Real(y - centre_y)/Real(bounds.h);
+ int64_t index = (x+target.w*y)*4;
- if (dx*dx + dy*dy <= 1.0)
+ if (dx*dx + dy*dy <= Real(1))
{
target.pixels[index+0] = 0;
target.pixels[index+1] = 0;
target.pixels[index+2] = 0;
target.pixels[index+3] = 255;
-
}
}
}
}
}
+Rect ObjectRenderer::CPURenderBounds(const Rect & bounds, const View & view, const CPURenderTarget & target)
+{
+ Rect result = view.TransformToViewCoords(bounds);
+ result.x *= Real(target.w);
+ result.y *= Real(target.h);
+ result.w *= Real(target.w);
+ result.h *= Real(target.h);
+ return result;
+}
+
+
+/**
+ * Bezier curve
+ * Not sure how to apply De'Casteljau, will just use a bunch of Bresnham lines for now.
+ */
+void BezierRenderer::RenderUsingCPU(const Objects & objects, const View & view, const CPURenderTarget & target, unsigned first_obj_id, unsigned last_obj_id)
+{
+ //Warn("Rendering Beziers on CPU. Things may explode.");
+ for (unsigned i = 0; i < m_indexes.size(); ++i)
+ {
+ if (m_indexes[i] < first_obj_id) continue;
+ if (m_indexes[i] >= last_obj_id) continue;
+ const Rect & bounds = objects.bounds[m_indexes[i]];
+ PixelBounds pix_bounds(CPURenderBounds(bounds,view,target));
+
+ Bezier control(objects.beziers[objects.data_indices[m_indexes[i]]].ToAbsolute(bounds),CPURenderBounds(Rect(0,0,1,1), view, target));
+ //Debug("%s -> %s via %s", objects.beziers[objects.data_indices[m_indexes[i]]].Str().c_str(), control.Str().c_str(), bounds.Str().c_str());
+ // Draw a rectangle around the bezier for debugging the bounds rectangle calculations
+ /*
+ ObjectRenderer::RenderLineOnCPU(pix_bounds.x, pix_bounds.y, pix_bounds.x+pix_bounds.w, pix_bounds.y, target, Colour(1,0,0,1));
+ ObjectRenderer::RenderLineOnCPU(pix_bounds.x, pix_bounds.y+pix_bounds.h, pix_bounds.x+pix_bounds.w, pix_bounds.y+pix_bounds.h, target, Colour(0,1,0,1));
+ ObjectRenderer::RenderLineOnCPU(pix_bounds.x, pix_bounds.y, pix_bounds.x, pix_bounds.y+pix_bounds.h, target, Colour(1,0,0,1));
+ ObjectRenderer::RenderLineOnCPU(pix_bounds.x+pix_bounds.w, pix_bounds.y, pix_bounds.x+pix_bounds.w, pix_bounds.y+pix_bounds.h, target, Colour(0,1,0,1));
+ */
+ // Draw lines between the control points for debugging
+ //ObjectRenderer::RenderLineOnCPU((int64_t)control.x0, (int64_t)control.y0, (int64_t)control.x1, (int64_t)control.y1,target);
+ //ObjectRenderer::RenderLineOnCPU((int64_t)control.x1, (int64_t)control.y1, (int64_t)control.x2, (int64_t)control.y2,target);
+
+
+
+ Real x[2]; Real y[2];
+ control.Evaluate(x[0], y[0], Real(0));
+ //Debug("target is (%lu, %lu)", target.w, target.h);
+ int64_t blen = 100;
+ //blen = min(max((int64_t)2, (int64_t)(target.w/view.GetBounds().w)), (int64_t)100);
+
+ Real invblen(1); invblen /= blen;
+ //Debug("Using %li lines, inverse %f", blen, Double(invblen));
+ for (int64_t j = 1; j <= blen; ++j)
+ {
+ control.Evaluate(x[j % 2],y[j % 2], invblen*j);
+ ObjectRenderer::RenderLineOnCPU((int64_t)Double(x[0]),(int64_t)Double(y[0]), (int64_t)Double(x[1]),(int64_t)Double(y[1]), target);
+ }
+
+ /*
+ Real u(0);
+ while (u < Real(1))
+ {
+ u += Real(1e-6);
+ Real x; Real y; control.Evaluate(x,y,u);
+ int64_t index = ((int64_t)x + (int64_t)y*target.w)*4;
+ if (index >= 0 && index < 4*(target.w*target.h))
+ {
+ target.pixels[index+0] = 0;
+ target.pixels[index+1] = 0;
+ target.pixels[index+2] = 0;
+ target.pixels[index+3] = 255;
+ }
+ }
+ */
+
+ }
+}
+
+void BezierRenderer::PrepareBezierGPUBuffer(const Objects& objects)
+{
+ m_bezier_coeffs.SetType(GraphicsBuffer::BufferTypeTexture);
+ m_bezier_coeffs.SetUsage(GraphicsBuffer::BufferUsageDynamicDraw);
+ m_bezier_coeffs.Resize(objects.beziers.size()*sizeof(GPUBezierCoeffs));
+ BufferBuilder<GPUBezierCoeffs> builder(m_bezier_coeffs.Map(false, true, true), m_bezier_coeffs.GetSize());
+
+
+ for (unsigned i = 0; i < objects.beziers.size(); ++i)
+ {
+ const Bezier & bez = objects.beziers[i];
+
+ GPUBezierCoeffs coeffs = {
+ Float(bez.x0), Float(bez.y0),
+ Float(bez.x1), Float(bez.y1),
+ Float(bez.x2), Float(bez.y2),
+ Float(bez.x3), Float(bez.y3)
+ };
+ builder.Add(coeffs);
+ }
+ m_bezier_coeffs.UnMap();
+ glGenTextures(1, &m_bezier_buffer_texture);
+ glBindTexture(GL_TEXTURE_BUFFER, m_bezier_buffer_texture);
+ glTexBuffer(GL_TEXTURE_BUFFER, GL_RG32F, m_bezier_coeffs.GetHandle());
+
+ m_bezier_ids.SetType(GraphicsBuffer::BufferTypeTexture);
+ m_bezier_ids.SetUsage(GraphicsBuffer::BufferUsageDynamicDraw);
+ m_bezier_ids.Upload(objects.data_indices.size() * sizeof(uint32_t), &objects.data_indices[0]);
+
+ glGenTextures(1, &m_bezier_id_buffer_texture);
+ glActiveTexture(GL_TEXTURE1);
+ glBindTexture(GL_TEXTURE_BUFFER, m_bezier_id_buffer_texture);
+ glTexBuffer(GL_TEXTURE_BUFFER, GL_R32I, m_bezier_ids.GetHandle());
+ glActiveTexture(GL_TEXTURE0);
+}
+
+void BezierRenderer::RenderUsingGPU(unsigned first_obj_id, unsigned last_obj_id)
+{
+ if (!m_shader_program.Valid())
+ Warn("Shader is invalid (objects are of type %d)", m_type);
+
+ // If we don't have anything to render, return.
+ if (first_obj_id == last_obj_id) return;
+ // If there are no objects of this type, return.
+ if (m_indexes.empty()) return;
+
+ unsigned first_index = 0;
+ while (m_indexes.size() > first_index && m_indexes[first_index] < first_obj_id) first_index ++;
+ unsigned last_index = first_index;
+ while (m_indexes.size() > last_index && m_indexes[last_index] < last_obj_id) last_index ++;
+
+ m_shader_program.Use();
+ glUniform1i(m_shader_program.GetUniformLocation("bezier_buffer_texture"), 0);
+ glUniform1i(m_shader_program.GetUniformLocation("bezier_id_buffer_texture"), 1);
+ m_ibo.Bind();
+ glDrawElements(GL_LINES, (last_index-first_index)*2, GL_UNSIGNED_INT, (GLvoid*)(2*first_index*sizeof(uint32_t)));
+}
+
+/**
+ * Render Group (shading)
+ */
+void GroupRenderer::RenderUsingCPU(const Objects & objects, const View & view, const CPURenderTarget & target, unsigned first_obj_id, unsigned last_obj_id)
+{
+ for (unsigned i = 0; i < m_indexes.size(); ++i)
+ {
+ if (m_indexes[i] < first_obj_id) continue;
+ if (m_indexes[i] >= last_obj_id) continue;
+
+
+ Rect bounds(CPURenderBounds(objects.bounds[m_indexes[i]], view, target));
+ PixelBounds pix_bounds(bounds);
+
+
+ Colour c(0.5,0.5,1,1);
+ // make the bounds just a little bit bigger
+ pix_bounds.x--;
+ pix_bounds.w++;
+ pix_bounds.y--;
+ pix_bounds.h++;
+ /*
+ ObjectRenderer::RenderLineOnCPU(pix_bounds.x, pix_bounds.y, pix_bounds.x+pix_bounds.w, pix_bounds.y, target, c);
+ ObjectRenderer::RenderLineOnCPU(pix_bounds.x, pix_bounds.y+pix_bounds.h, pix_bounds.x+pix_bounds.w, pix_bounds.y+pix_bounds.h, target, c);
+ ObjectRenderer::RenderLineOnCPU(pix_bounds.x, pix_bounds.y, pix_bounds.x, pix_bounds.y+pix_bounds.h, target, c);
+ ObjectRenderer::RenderLineOnCPU(pix_bounds.x+pix_bounds.w, pix_bounds.y, pix_bounds.x+pix_bounds.w, pix_bounds.y+pix_bounds.h, target, c);
+ */
+ // Attempt to shade the region
+ // Assumes the outline has been drawn first...
+ for (int64_t y = max((int64_t)0, pix_bounds.y); y <= min(pix_bounds.y+pix_bounds.h, target.h-1); ++y)
+ {
+ bool inside = false;
+ bool online = false;
+ for (int64_t x = max((int64_t)0, pix_bounds.x); x <= min(pix_bounds.x+pix_bounds.w, target.w-1); ++x)
+ {
+ int64_t index = (x+target.w*y)*4;
+ if (target.pixels[index+0] == 0 && target.pixels[index+1] == 0 && target.pixels[index+2] == 0 && target.pixels[index+3] == 255)
+ {
+ online = true;
+ continue;
+ }
+ else if (online)
+ {
+ inside = !inside;
+ online = false;
+ }
+
+ if (inside)
+ {
+ target.pixels[index+0] = c.r*255;
+ target.pixels[index+1] = c.g*255;
+ target.pixels[index+2] = c.b*255;
+ target.pixels[index+3] = c.a*255;
+ }
+ }
+ }
+
+
+ }
+}
/**
* For debug, save pixels to bitmap
SDL_FreeSurface(surf);
}
+/**
+ * Bresenham's lines
+ */
+void ObjectRenderer::RenderLineOnCPU(int64_t x0, int64_t y0, int64_t x1, int64_t y1, const CPURenderTarget & target, const Colour & colour, bool transpose)
+{
+ int64_t dx = x1 - x0;
+ int64_t dy = y1 - y0;
+ bool neg_m = (dy*dx < 0);
+ dy = abs(dy);
+ dx = abs(dx);
+
+ // If positive slope > 1, just swap x and y
+ if (dy > dx)
+ {
+ RenderLineOnCPU(y0,x0,y1,x1,target,colour,!transpose);
+ return;
+ }
+
+ int64_t two_dy = 2*dy;
+ int64_t p = two_dy - dx;
+ int64_t two_dxdy = 2*(dy-dx);
+ int64_t x; int64_t y; int64_t x_end;
+ int64_t width = (transpose ? target.h : target.w);
+ int64_t height = (transpose ? target.w : target.h);
+
+ uint8_t rgba[4];
+ rgba[0] = 255*colour.r;
+ rgba[1] = 255*colour.g;
+ rgba[2] = 255*colour.b;
+ rgba[3] = 255*colour.a;
+
+ if (x0 > x1)
+ {
+ x = x1;
+ y = y1;
+ x_end = x0;
+ }
+ else
+ {
+ x = x0;
+ y = y0;
+ x_end = x1;
+ }
+
+ if (x < 0)
+ {
+ if (x_end < 0) return;
+ y = (neg_m ? y - (dy*-x)/dx : y + (dy*-x)/dx);
+ x = 0;
+ }
+
+ if (x_end > width)
+ {
+ if (x > width) return;
+ x_end = width-1;
+ }
+
+ // TODO: Avoid extra inner conditionals
+ do
+ {
+ if (x >= 0 && x < width && y >= 0 && y < height)
+ {
+ int64_t index = (transpose ? (y + x*target.w)*4 : (x + y*target.w)*4);
+ for (int i = 0; i < 4; ++i)
+ target.pixels[index+i] = rgba[i];
+ }
+
+ if (p < 0)
+ p += two_dy;
+ else
+ {
+ if (neg_m) --y; else ++y;
+ p += two_dxdy;
+ }
+ } while (++x <= x_end);
+}
+
}