When changing from He to H2, you should not expect any improvement in the resolution. The only case when the resolution improvement is possible, it is when the improvement is possible with both gases, i.e. when the original analysis with He was not optimized for the best resolution. The key here is that carrier gas does not affect the best possible resolution.
What's good about H2 is that it is the fastest carrier gas.
In a method translated from He to H2, the flow rate of H2 should be 25% higher than was the flow rate of He. The temperature program should be translated accordingly (see below). The exact changes in GC-MS are different from those in GC-FID. The former ones are easier to predict.
GC-MS: Method translation from He to H2 reduces the analysis time by 40%. In a temperature-programmed analysis, the hold-time for each temperature should be reduced by 40%, and each heating rate should be increased by 68% (as 1/0.6 = 1.667). That's all.
The changes in GC-FID are more difficult to predict in the short note like this because they depend on column diameter and length. Method translation from He to H2 reduces the analysis time by 20% to 40%. To figure out the changes in the temperature program, you can use a GC Method Translation freeware available on line (see below), or do the following. (a) Measure the void time (tMHe) in the original analysis with He. (b) Measure the void time (tMH2) in analysis with H2. (c) Calculate the time reduction factor (TRF) as TRF = tMH2/tMHe. (d) Calculate the time (tHoldH2) of each temperature-hold in the method with H2 as tHoldH2 = TRF*tHoldHe (tHoldHe = temperature-hold time with He). (e) Calculate each heating rate (RateH2) in analysis with H2 as RateH2 = RateHe/TRF.
OR, to translate any method (GC-MS, GC-FID, etc.) get GC Method Translation freeware. I think that
http://www.restek.com/ezgc-mtfc is the best. In the "Results" choose "Solve for" = "Translate". You can also try "Solve for" = "Speed". Most likely, it will recommend the parameters for the best resolution-speed tradeoff. Don't use "Solve for" = "Efficiency". This option is practically worthless, and should not be there.
Additional comment. If the extra-column peak broadening in your system is significant (due to insufficiently sharp sample introduction, insufficiently fast detector, etc.) then switching from helium to hydrogen might cause some loss in the resolution because the faster is the analysis the more sensitive it is to the extra-column peak broadening.