Electric vehicles (EVs) have great potential for implementing demand response and regulating power loads. However, existing methods mainly rely on price control and load baselines, with single response objectives, and do not fully consider user-specific needs and system uncertainties, making it difficult to achieve desired user incentives. To address this, an EV demand response optimization method based on electricity-carbon load baselines is proposed. First, an EV node electricity-carbon coupling demand response model is developed, considering parking demand variations, charging randomness, and dynamic carbon emission factors. Then, a customer directrix load dual-objective game rolling optimization framework is introduced, incorporating a day-ahead and intra-day rolling optimization model and a dual-objective leader-follower game between the distribution network and EV users to set more precise customer directrix load. A joint game optimization algorithm is proposed to reduce computational complexity and iteration time while ensuring global optimality and overall solution effectiveness. Case studies show that the method optimizes electricity-carbon balance in the distribution network, effectively reduces EV user response difficulty, ensures mutual benefits, and provide a reference for accurate baseline setting.