I built one of these and it's an excellent re-purposing idea!  Mine is PAPR however and I find it easier to breathe with no effort due to the pressurization.  My ebay mask has independent intake and exhaust - I had to plug the dual exhaust tubes just above the one-way valving after I cut the end of the top tube (holding the ball...).  I'm using an ebay computer fan Delta BFB1012M and a canister HEPA filter glued onto it, with a light plastic washing machine discharge hose to convey the air up and into the mask.  Estimate I'll get better than 4 hours service from 4 X 18650 batteries.  Adapting the hose to the mask intake (odd shape) and fan exhaust (odd shape...) was the hardest part.

Thanks Joe yes I had bought some fans and a 18650 batteries with that idea in mind, but the lung-powered mask really took off. I appreciate that for some people the breathing effort might be a bit much with the passive design. It's worse with the HEPA filter in place because the filter has more air resistance.

What is the size of an N95 filter in microns compared to the pollen filter you are using? Is the pollen filter adequate to prevent the virus from entering? Great idea by the way!

The choice of filter material is yours of course. I would go for HEPA filters if possible  for maximum protection e.g. the  Bosch HEPA cabin filter spec states:  "efficiency of 99.97% at 0.3 microns* . Most of the exhaled particles you'll want to catch are cough droplets down to around 0.1 microns, and most range between 0.1 and 10um. N95 spec states 95% at 0.3 micron, although the genuine 3M ones exceed that spec by a little, I think I read 97%. So if you choose the right filter material, even meant for cars, you can exceed the baseline N95 spec, you are heading for the N100 spec. As was also pointed out below the filter is out of the "firing line" of the splatter, so may fare better for that reason too. A standard pollen filter will be trapping particles down to 10um and then dropping off below that. But you can always double or treble the thickness to increase filtration

Thank you for the answer and nice job on this hack.

Be careful that the valve at the front of the mouth closes quickly - some fit testing suggests that this might leak on initial inspiration. Sealing it is not a good idea as CO2 levels are likely to rise quite quickly. 

Thanks John yes -  next upgrades added to the project are to tape more filter material over each valve, there are two between the eye area and the mouth, and the exhale valves too. Most of the exhale in this model goes up around the face through air ducts (to keep moist air away from the eye area which would mist it up) and out of the filter on the top.  In normal use it would go out of the snorkel attachment, so you can flush the water out before breathing in. But yes, the front valve was a concern and I've addressed that too.  I have posted up more pictures of the type of model you should ideally be looking for, and the features which I think are good, and the internal filter upgrades.

Nice work! Best thing is, the air inlet is not directly where patients would cough/sneeze against!

Yeah I did that on purpose! ;^) Actually you could take a tube off the inlet down to a filter canister (even under your gown if in ICU) which would improve that aspect by a factor of 10 or more. The other great thing is that the mask is powered by your own lungs - no fans, no power pack to run down.

Thanks Dan. Like all the best projects it was hardly any work  - so anyone can do it!

I added this to the COVID-19 list. Nice work