To make a standard curve you have to use at least 3 reference genes with serial dilutions of each gene. This step is called validation. The gene could be used as reference should have stable expression all the time. You should search for reference genes of the species you working on. E.g. Reference genes used for bacteria Strep. could be gyrA, glnA, shdA.
Why do you want to run absolute AND relative quantification? You can choose one or the other. If you decide to run a absolute quantification (which can be extra hard since you are interested in several genes) you'll need to amplify, purify and quantify all genes in order to a set a dilution curve. If I were you I would go for the relative quantification using a good reference (you should check the literature and also experimentally).
First you decide what you are want to do absolute or relative quantification.
You need to doing standard curve for calculating efficiency of reaction and you can doing through diluting one of your positive sample many time for example at least 5 dilutions (1/10, 1/100. 1/1000, 1/10000, 1/100000) for both your gene of interest and reference gene but independently (after choose the appropriate reference gene)
You can used different dilutions of your reference template to generate the standard curve. Take the sample of each dilution in replicate. Depending on standard curve you can quantify your unknown templates and use can such quantified templated for relative quantification.