Update SSBO in Compute shader - opengl

Question

I am currently trying to update a SSBO linked/bound to a Computeshader. Doing it this way, I only write the first 32byte into the out_picture, because i only memcpy that many (sizeof(pstruct)).
Computeshader:
#version 440 core
struct Pstruct{
float picture[1920*1080*3];
float factor;
};
layout(std430, binding = 0) buffer Result{
float out_picture[];
};
layout(std430, binding = 1) buffer In_p1{
Pstruct in_p1;
};
layout(local_size_x = 1000) in;
void main() {
out_picture[gl_GlobalInvocationID.x] = out_picture[gl_GlobalInvocationID.x] +
in_p1.picture[gl_GlobalInvocationID.x] * in_p1.factor;
}
GLSL:
struct Pstruct{
std::vector<float> picture;
float factor;
};
Pstruct tmp;
tmp.factor = 1.0f;
for(int i = 0; i < getNUM_PIX(); i++){
tmp.picture.push_back(5.0f);
}
SSBO ssbo;
glGenBuffers(1, &ssbo.handle);
glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 1, ssbo.handle);
glBufferData(GL_SHADER_STORAGE_BUFFER, (getNUM_PIX() + 1) * sizeof(float), NULL, GL_DYNAMIC_DRAW);
...
glBindBuffer(GL_SHADER_STORAGE_BUFFER, ssbo.handle);
Pstruct* ptr = (Pstruct *) glMapBuffer(GL_SHADER_STORAGE_BUFFER, GL_WRITE_ONLY);
memcpy(ptr, &pstruct, sizeof(pstruct));
glUnmapBuffer(GL_SHADER_STORAGE_BUFFER);
...
glUseProgram(program);
glDispatchCompute(getNUM_PIX() / getWORK_GROUP_SIZE(), 1, 1);
glMemoryBarrier(GL_SHADER_STORAGE_BARRIER_BIT);
How can I copy both my picture array and my float factor at the same time?
Do I have to split the memcpy call into array and float? and if yes how? I can copy the first part, but I am not allowed to add an offset to the ptr.

Answer 1

First of all,
float picture[1920*1080*3];
clearly should be either a texture (you're only reading from it anyway) or at least an image.
Second:
struct Pstruct{
std::vector<float> picture;
float factor;
};
This definition does not match the definition in your shader in any way. The std::vector object will just be a meta object internally managing the data storage used by the vector. memcpy that to a GL buffer and passing that to the GPU does not make sense at all.
The correct approach would be to either copy the contents of that vector separately into the appropriate places inside the buffer, or to just us a struct definition on your client side which actually matches the one you're using in the shader (and taking all the rules of std430 into account). But, as my first point already was, the correct solution here is most likely to use a texture or image object instead.