I have also considering to make own pyramid absorber. I would like to have compact chamber for antenna measurements at around 900MHz. I think I would need about 300mm thick pyramid absorber. Making own absorber probably does not make much sense, but it is interesting topic. High quality absorbers from well known manufacturer are very expensive. Cheap absorbers from unknown manufacturer can be good, but there is risk that you don't get what you expect.
There is rather detailed document about manufacturing process (page 7) from ETS Lindgren:
http://www.ets-lindgren.com/sites/etsauthor/General_Brochures/Top%2010%20Anechoic%20Absorber%20Considerations.pdf It looks like fairly simple. I did quick test from old mattress. I was using electric paint from Bare Conductive. I mixed some water to the paint. Then I soak the piece of mattress to this solution. To speed up the process I was making curing at elevated temperature. Outcome looked pretty good and feels and looks like commercial absorber. I didn't perform any RF testing, but the foam had similar DC resistance than commercial absorber measured with multimeter.
Cutting pyramids is a bit challenging with DIY tools. My absorber foam was shrinking a bit after curing, so it need to be cut from all sides. I think some kind of special scroll saw would be needed to cut shapes from the foam block.
Another challenge is to find thick mattress foam. I think 150mm foam is easy to find. I'm not sure, what would happen if layers of foam is glued together. I guess some solvent based glue would work in soaking process. Glue layer might have some impact to RF properties.
Final issue would be the characterization of own absorber. Probably the easiest way is to use the NRL arch method, which is basically bistatic measurement. The reflection of absorber compared with the perfect electric conductor. It would take big effort to make prototype sample for the measurement.
It would help if there would be proven formula of the soaking solution and curing process. I'm not sure if some iron oxide or graphite or some other material should be added to formula?