Is this GPU video transcoding project feasible?

Question

Recently I was approached by a guy who wants to do video transcoding using the GPU. He basically wants me to create him an application that he can sell or gain revenue from advertising.

Now he has basically asked me to tell me what I can achieve with 5000 US dollars of pay. Now, I am a graduate student and won an award for my final thesis which was based on evolutionary algorithms using the GPU. So I know about GPGPU progrmaming (which is why we got in contact in the first place), but I haven't got any indepth knowledge of video compression and transcoding and the like.

So I'm at a bit of a loss, I would like the project but I don't want to sell myself short and I don't want to promise the unachievable. Considering he wants a product similar to the competition (badaboom, avivo) and I imagine they are teams of people with much more knowledge about the topic than I, I am somewhat concerned.

So I guess my question is how would you respond if you were in my situation, and how difficult do you think a project (writing a GPU-based video transcoder) would be for an able but not expert in this field person like me?

Thanks in advance

p.s. he wants it to be cross-platform

Answer 1

You seem to be underselling yourself. I guess the rate for a GPGPU programming expert in the US should be around $100/hr, if not more. That's 50 hours or roughly 6 working days, including testing, debugging and whatever else you need to do. That's an awful short amount of time, even if your rate is less.

Answer 2

AMD provides a gpu-based transcoder that's a free download. They call it AVIVO.

If good products are available as free downloads, your client will have trouble selling a product, so be sure and get paid in advance.