Rav says that the keren is evaluated based on the value at the time of geneiva, but the k'nas is evaluated based on the time of the ruling of beis din. The gemara explains that in a situation where the object either appreciates in value or depreciates in value to do the investment or mistreatment of the ganev, all would agree that the ganev reimburses based on the value at the time he stole it because the owner should not be entitled to appreciation due to the ganev's investment, nor should he be penalized by depreciation due to the ganev's destruction.
The gemara explains that Rav's statment applies only to a case of a price drop. Being that we are speaking about a case where he physically destroys the animal (such as shechita for which he pays 4 or 5), had the price increased from the time of the geneiva, he should pay even the keren based on the value at the time he slaughtered the animal. Rav's distinction is limited to a case where the price drops between the geneiva and ha'amada b'din. When it was originally worth more and is now worth less due to a price drop, the keren is evaluated based on the time of the damage but the k'nas is evaluted based on ha'amada b'din. Even this is later qualified by the gemara 66a - Rashi learns that when he pays back the same type of item that he stole i.e. sheep, he pays keren and k'nas based on the time that he stole, but when he pays cash that is where rav said that the k'nas is paid based on ha'amada b'din. R"I (in tosafos) learns that any improvement or deterioration of the animal whether or not it was caused by him is paid based on the price at the time of the sale both the keren and the k'nas. But for any change in price the k'nas is evaluated based on the market value at the ha'amada b'din [but the keren is still evaluated based on the time of geneiva].
The p'nei yehoshua explains that the concept of k'nas being paid based on the price of ha'amda b'din is very logical and should be assumed, since the k'nas is only realized at the time that the beis din rules on it. Since until that point the ganev had the ablity to exempt himself by admitting, it makes sense to say that he pays based on the ha'amada b'din price. However, Tosafos (d.h. gufa) holds that the chiddush of rav that we need a special pasuk to teach, אחייה לקרן כעין שגנב, is not for the keren, but rather for the implication that k'nas is paid based on ha'amada b'din. The question is, given the logic of the pnei yehoshua, this should also be obvious and not require a special pasuk? Tosafos seems to go lishatasam (d.h. alma) that the k'nas should logically follow the keren, so that even if the k'nas should logically be evaluated based on ha'amada b'din, it should follow the keren to be evaluated based on the time of geneiva. We therefore require a gezeiras hakasuv that ONLY keren pays based on the value at the time of geneiva, but k'nas pays based on the value at the time of ha'amada b'din (which is a logical distinction, although without a pasuk we would have tied the k'nas to the keren).
Rashi 65b d.h. d'kfeila, seems to argue on Tosafos and holds that according to rav we do require a pasuk to teach the chiddush of keren being paid based on the value at the time of the geneiva. Rashi implies that both the aspect of keren being evaluated based on the time of geneiva, and the aspect of k'nas being evaluated based on the value at the time of ha'amada b'din are part of the chiddush of Rav.