Hello?
ID AlphaDol.
I'd like to ask you a question.
As shown in the attached picture, the results of the CVPR paper are slightly different from those simulated by code.
The CVPR paper seems to have a more similar result to Gogh's picture.
So my question is, which part of the code you provided (or which parameters) should I touch to adjust this part? For example, can I adjust the loss ratio in the code below?
I'd appreciate it if you could give me an insight into this. ^^
In the simulation results on the left, the lower left side has a bit of red light like noise, but there is no red part like this in the original Gogh drawing, but I want to know why this part is created in videos that are not related to noise.
Thank you.
def entire_loss(gens,cons,stys):
sl,cl = 0,0 # initialization
for gen,con,sty in zip(gens,cons,stys):
cl += 0.2*C_loss(con,gen) # 5 layers 20% allocation
sl += 0.2*S_loss(sty,gen)
Total = 10*cl + 40*sl # Rates from bibliography <-- this part?
return total,sl,cl