transformation on variables

I run simple linear regression, and saw that my residuals not distributed normally.
my dependent variable histogram was as log-normal, i did a log-normal transformation to him. After that i run linear regression with log(dependent variable), and got results.
I attached the results. what do you think? the kind of transformation was made right?
Now the results is much better?
i did the same regression for different data, maybe you can recommend me how improve it?
thank you